Tutorial_dp-GAN.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<script>\n",
       "code_show=true; \n",
       "function code_toggle() {\n",
       " if (code_show){\n",
       " $('div.input').hide();\n",
       " } else {\n",
       " $('div.input').show();\n",
       " }\n",
       " code_show = !code_show\n",
       "} \n",
       "$( document ).ready(code_toggle);\n",
       "</script>\n",
       "<form action=\"javascript:code_toggle()\"><input type=\"submit\" value=\"Click here to toggle on/off the raw code.\"></form>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import HTML\n",
    "\n",
    "HTML('''<script>\n",
    "code_show=true; \n",
    "function code_toggle() {\n",
    " if (code_show){\n",
    " $('div.input').hide();\n",
    " } else {\n",
    " $('div.input').show();\n",
    " }\n",
    " code_show = !code_show\n",
    "} \n",
    "$( document ).ready(code_toggle);\n",
    "</script>\n",
    "<form action=\"javascript:code_toggle()\"><input type=\"submit\" value=\"Click here to toggle on/off the raw code.\"></form>''')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# How to  generate new data using Generative Adversarial Networks\n",
    "# in a differential private mode \n",
    "\n",
    "   \n",
    "#   \n",
    "#### ***by Lyudmyla Dymytrova, Lorenzo Frigerio, and Anderson Santana de Oliveira***\n",
    "August 2019"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###    \n",
    "## Introduction\n",
    "\n",
    "   This tutorial shows how to generate a new (fake) dataset using Generative Adversarial Network as reported in  Frigerio L., Santana de Oliveira A., Gomez L. & Duverger P. (2019) Differentially private GANs for time series, continuous and discrete Open Data) (http://arxiv.org/abs/1901.02477). With the exception that we use out of the box, AutoML classifier for checking the quality of the generated data. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING: Logging before flag parsing goes to stderr.\n",
      "W0816 14:43:42.869918 35388 deprecation_wrapper.py:119] From C:\\Users\\i061767\\GitHub\\testbed\\differential_privacy\\dp_sgd\\dp_optimizer\\dp_optimizer.py:27: The name tf.train.AdamOptimizer is deprecated. Please use tf.compat.v1.train.AdamOptimizer instead.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Load modules\n",
    "import os, sys\n",
    "\n",
    "\n",
    "import random\n",
    "import matplotlib\n",
    "matplotlib.use('Agg')\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import csv\n",
    "import time\n",
    "import pickle \n",
    "\n",
    "import tflib as lib\n",
    "import tflib.ops.linear\n",
    "import tflib.plot\n",
    "import tflib.inception_score\n",
    "\n",
    "from differential_privacy.dp_sgd.dp_optimizer import dp_optimizer\n",
    "from differential_privacy.dp_sgd.dp_optimizer import sanitizer\n",
    "from differential_privacy.privacy_accountant.tf import accountant\n",
    "\n",
    "from discretedata_manager import discretedata_manager\n",
    "from models import Generator, Discriminator\n",
    "\n",
    "from scipy.stats import wasserstein_distance, chi2_contingency\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import normalize\n",
    "from sklearn.metrics import accuracy_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Checking whether there is an H2O instance running at http://localhost:54321 ..... not found.\n",
      "Attempting to start a local H2O server...\n",
      "; Java HotSpot(TM) 64-Bit Server VM (build 25.212-b10, mixed mode)\n",
      "  Starting server from C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\backend\\bin\\h2o.jar\n",
      "  Ice root: C:\\Users\\i061767\\AppData\\Local\\Temp\\tmpiw_hh7mr\n",
      "  JVM stdout: C:\\Users\\i061767\\AppData\\Local\\Temp\\tmpiw_hh7mr\\h2o_i061767_started_from_python.out\n",
      "  JVM stderr: C:\\Users\\i061767\\AppData\\Local\\Temp\\tmpiw_hh7mr\\h2o_i061767_started_from_python.err\n",
      "  Server is running at http://127.0.0.1:54321\n",
      "Connecting to H2O server at http://127.0.0.1:54321 ... successful.\n",
      "Warning: Your H2O cluster version is too old (3 months and 9 days)! Please download and install the latest version from http://h2o.ai/download/\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td>H2O cluster uptime:</td>\n",
       "<td>06 secs</td></tr>\n",
       "<tr><td>H2O cluster timezone:</td>\n",
       "<td>Europe/Paris</td></tr>\n",
       "<tr><td>H2O data parsing timezone:</td>\n",
       "<td>UTC</td></tr>\n",
       "<tr><td>H2O cluster version:</td>\n",
       "<td>3.24.0.3</td></tr>\n",
       "<tr><td>H2O cluster version age:</td>\n",
       "<td>3 months and 9 days !!!</td></tr>\n",
       "<tr><td>H2O cluster name:</td>\n",
       "<td>H2O_from_python_i061767_iqxtb0</td></tr>\n",
       "<tr><td>H2O cluster total nodes:</td>\n",
       "<td>1</td></tr>\n",
       "<tr><td>H2O cluster free memory:</td>\n",
       "<td>3.521 Gb</td></tr>\n",
       "<tr><td>H2O cluster total cores:</td>\n",
       "<td>0</td></tr>\n",
       "<tr><td>H2O cluster allowed cores:</td>\n",
       "<td>0</td></tr>\n",
       "<tr><td>H2O cluster status:</td>\n",
       "<td>accepting new members, healthy</td></tr>\n",
       "<tr><td>H2O connection url:</td>\n",
       "<td>http://127.0.0.1:54321</td></tr>\n",
       "<tr><td>H2O connection proxy:</td>\n",
       "<td>None</td></tr>\n",
       "<tr><td>H2O internal security:</td>\n",
       "<td>False</td></tr>\n",
       "<tr><td>H2O API Extensions:</td>\n",
       "<td>Amazon S3, Algos, AutoML, Core V3, Core V4</td></tr>\n",
       "<tr><td>Python version:</td>\n",
       "<td>3.6.3 final</td></tr></table></div>"
      ],
      "text/plain": [
       "--------------------------  ------------------------------------------\n",
       "H2O cluster uptime:         06 secs\n",
       "H2O cluster timezone:       Europe/Paris\n",
       "H2O data parsing timezone:  UTC\n",
       "H2O cluster version:        3.24.0.3\n",
       "H2O cluster version age:    3 months and 9 days !!!\n",
       "H2O cluster name:           H2O_from_python_i061767_iqxtb0\n",
       "H2O cluster total nodes:    1\n",
       "H2O cluster free memory:    3.521 Gb\n",
       "H2O cluster total cores:    0\n",
       "H2O cluster allowed cores:  0\n",
       "H2O cluster status:         accepting new members, healthy\n",
       "H2O connection url:         http://127.0.0.1:54321\n",
       "H2O connection proxy:\n",
       "H2O internal security:      False\n",
       "H2O API Extensions:         Amazon S3, Algos, AutoML, Core V3, Core V4\n",
       "Python version:             3.6.3 final\n",
       "--------------------------  ------------------------------------------"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Initialize H2O.api for AutoML \n",
    "import h2o\n",
    "from h2o.automl import H2OAutoML\n",
    "h2o.init()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>age</th>\n",
       "      <th>workclass</th>\n",
       "      <th>fnlwgt</th>\n",
       "      <th>education</th>\n",
       "      <th>education-num</th>\n",
       "      <th>marital-status</th>\n",
       "      <th>occupation</th>\n",
       "      <th>relationship</th>\n",
       "      <th>race</th>\n",
       "      <th>sex</th>\n",
       "      <th>capital-gain</th>\n",
       "      <th>capital-loss</th>\n",
       "      <th>hours-per-week</th>\n",
       "      <th>native-country</th>\n",
       "      <th>output</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>39</td>\n",
       "      <td>State-gov</td>\n",
       "      <td>77516</td>\n",
       "      <td>Bachelors</td>\n",
       "      <td>13</td>\n",
       "      <td>Never-married</td>\n",
       "      <td>Adm-clerical</td>\n",
       "      <td>Not-in-family</td>\n",
       "      <td>White</td>\n",
       "      <td>Male</td>\n",
       "      <td>2174</td>\n",
       "      <td>0</td>\n",
       "      <td>40</td>\n",
       "      <td>United-States</td>\n",
       "      <td>Fal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>50</td>\n",
       "      <td>Self-emp-not-inc</td>\n",
       "      <td>83311</td>\n",
       "      <td>Bachelors</td>\n",
       "      <td>13</td>\n",
       "      <td>Married-civ-spouse</td>\n",
       "      <td>Exec-managerial</td>\n",
       "      <td>Husband</td>\n",
       "      <td>White</td>\n",
       "      <td>Male</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>13</td>\n",
       "      <td>United-States</td>\n",
       "      <td>Fal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>38</td>\n",
       "      <td>Private</td>\n",
       "      <td>215646</td>\n",
       "      <td>HS-grad</td>\n",
       "      <td>9</td>\n",
       "      <td>Divorced</td>\n",
       "      <td>Handlers-cleaners</td>\n",
       "      <td>Not-in-family</td>\n",
       "      <td>White</td>\n",
       "      <td>Male</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>40</td>\n",
       "      <td>United-States</td>\n",
       "      <td>Fal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>53</td>\n",
       "      <td>Private</td>\n",
       "      <td>234721</td>\n",
       "      <td>11th</td>\n",
       "      <td>7</td>\n",
       "      <td>Married-civ-spouse</td>\n",
       "      <td>Handlers-cleaners</td>\n",
       "      <td>Husband</td>\n",
       "      <td>Black</td>\n",
       "      <td>Male</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>40</td>\n",
       "      <td>United-States</td>\n",
       "      <td>Fal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>28</td>\n",
       "      <td>Private</td>\n",
       "      <td>338409</td>\n",
       "      <td>Bachelors</td>\n",
       "      <td>13</td>\n",
       "      <td>Married-civ-spouse</td>\n",
       "      <td>Prof-specialty</td>\n",
       "      <td>Wife</td>\n",
       "      <td>Black</td>\n",
       "      <td>Female</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>40</td>\n",
       "      <td>Cuba</td>\n",
       "      <td>Fal</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   age         workclass  fnlwgt  education  education-num  \\\n",
       "0   39         State-gov   77516  Bachelors             13   \n",
       "1   50  Self-emp-not-inc   83311  Bachelors             13   \n",
       "2   38           Private  215646    HS-grad              9   \n",
       "3   53           Private  234721       11th              7   \n",
       "4   28           Private  338409  Bachelors             13   \n",
       "\n",
       "       marital-status         occupation   relationship   race     sex  \\\n",
       "0       Never-married       Adm-clerical  Not-in-family  White    Male   \n",
       "1  Married-civ-spouse    Exec-managerial        Husband  White    Male   \n",
       "2            Divorced  Handlers-cleaners  Not-in-family  White    Male   \n",
       "3  Married-civ-spouse  Handlers-cleaners        Husband  Black    Male   \n",
       "4  Married-civ-spouse     Prof-specialty           Wife  Black  Female   \n",
       "\n",
       "   capital-gain  capital-loss  hours-per-week native-country output  \n",
       "0          2174             0              40  United-States    Fal  \n",
       "1             0             0              13  United-States    Fal  \n",
       "2             0             0              40  United-States    Fal  \n",
       "3             0             0              40  United-States    Fal  \n",
       "4             0             0              40           Cuba    Fal  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "# Load dataset for training\n",
    "adult = pd.read_csv(\"./dataset/adult.csv\",delimiter=',',\n",
    "                    skipinitialspace = True)\n",
    "adult.head()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data preprocessing\n",
    "\n",
    "The first step is a preparation of data to a specific format for training WGAN. The input data set must be in a format of Pandas DataFrame. Data can be labelled or unlabeled. \n",
    "- **Missing values** is an issue only for classifier and do not impact the WGAN itself, so the next step might be skipped for unlabeled data. If dataset contains missing values, by default columns with more than 50% of NaN will be removed. The number of permitted missing values in a column can be reduced by user (an argument na_percent). If the number of permitted missing values in a column is less than a given threshold, each missing value will be replaced by a string that combines ‘missing’ and the column name where this missing value occurs. This procedure allows to distinguish between NaN values that occur in different columns. \n",
    "- So far only **categorical** features are implemented (the algorithm automatically removes all numerical features except a target column). If data labelled, a target column must be binary due to H2O AutoML does not have implemented multilabel classification yet??? A **target column** can be categorical (with two levels) or numerical. \n",
    "- The algorithm performs a random subsampling of majority class if data are **imbalanced**. Within the preprocessing step a function ‘prepare_data’ performs a random shuffling and **splitting data** into test (33% of data) and train (67%) datasets preserving a class balance for the target column in both datasets.  \n",
    "- There is no train/test split for an **unlabelled** data. The entire input dataset is used for WGAN training.\n",
    "- **The output** contains 4 instances, i.e. train_data (data frame), test_data (data frame), variables (list of strings) and target (string) that are used at next steps to train WGAN.\n",
    "\n",
    "```adult = pd.read_csv(\"./dataset/adult.csv\", delimiter=',',  skipinitialspace = True)\n",
    "prepData = prepare_data(adult, target='output')\n",
    "prepData.test_data.shape\n",
    "prepData.target\n",
    "prepData.variables```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Data preprocessing with train/test splitting \n",
    "\n",
    "class prepare_data(object):\n",
    "    def __init__(self, data, target = None, na_percent = 0.5):\n",
    "        self.data = data\n",
    "        self.target = target\n",
    "        self.na_percent = na_percent\n",
    "        self.train_data, self.test_data, self.variables  = self.prepare_discrete_data()\n",
    " \n",
    "    def prepare_discrete_data(self):\n",
    "        TextData = self.data.select_dtypes(include = ['object']).copy()\n",
    "        discrete_cols = TextData.columns\n",
    "        NewData = TextData[discrete_cols]\n",
    "        # Check if there are NaN values \n",
    "        if NewData.isnull().values.any() == True:\n",
    "            # Remove columns with more than xx% of NaN (xx% is user-defined or 0.5 by default)\n",
    "            NewData.dropna(thresh = NewData.shape[0]*(1-self.na_percent), how ='all', axis = 1, inplace = True)\n",
    "            # In rest of columns replace NaN by a string 'missing_column_name'\n",
    "            for col, name in enumerate(NewData.columns):\n",
    "                values = {name: str(\"missing \"+ NewData.columns[col])}\n",
    "                NewData.fillna(value = values, inplace = True)\n",
    "        # If data is unlabelled, no preprocessing\n",
    "        if self.target == None:\n",
    "            train_data = NewData\n",
    "            test_data = pd.DataFrame()\n",
    "            print('WARNING: no test data for unlabelled dataset')\n",
    "            return train_data, test_data, discrete_cols\n",
    "        else:\n",
    "            # If data is labelled, check a format of the target variable\n",
    "            if self.data[self.target].dtype != object:\n",
    "                NewData = pd.concat((TextData, self.data[self.target]), axis = 1)\n",
    "                discrete_cols = NewData.columns\n",
    "            pos = NewData.loc[NewData[self.target] == NewData[self.target].unique()[0]]\n",
    "            neg = NewData.loc[NewData[self.target] == NewData[self.target].unique()[1]]\n",
    "            # Check if classes are balanced\n",
    "            if len(pos) == len(neg):\n",
    "                train_data, test_data = train_test_split(NewData, test_size = 0.33, random_state = 1, shuffle = True,\n",
    "                                                             stratify = NewData[self.target])\n",
    "                return train_data, test_data, discrete_cols\n",
    "            # If classes are unbalanced, randomly subsample majority class\n",
    "            elif len(pos) < len(neg):\n",
    "                neg = neg.sample(len(pos))\n",
    "                data = pd.concat((pos,neg),axis=0)\n",
    "                train_data, test_data = train_test_split(data, test_size = 0.33, random_state = 1, shuffle = True,\n",
    "                                                             stratify = data[self.target])\n",
    "                return train_data, test_data, discrete_cols\n",
    "            else:\n",
    "                pos = pos.sample(len(neg))\n",
    "                data = pd.concat((pos,neg),axis=0)\n",
    "                train_data, test_data = train_test_split(data, test_size = 0.33, random_state = 1, shuffle = True,\n",
    "                                                             stratify = data[self.target])\n",
    "                return train_data, test_data, discrete_cols"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5176, 9)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# ADULT DATASET:  Data preprocessing\n",
    "adult_pr = prepare_data(adult, target='output')\n",
    "adult_pr.test_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Tru    5253\n",
       "Fal    5253\n",
       "Name: output, dtype: int64"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "adult_pr.train_data.output.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Tuning WGAN hyperparameters\n",
    "\n",
    "The next important step is a choice of hyperparameters for training WGAN. This can be done at one step using a function ‘WGAN_params’. \n",
    "- **privacy**: boolean, True or False. The user can choose to train WGAN in a differential private or public mode. \n",
    "- **batch_size**: the number of samples n that are processed by WGAN at once. The generator generates n fake samples (batch) that are further compared by discriminator to a randomly sampled n samples (batch) of real training data. After processing a batch, generator parameters are updating, and a new set of n samples (batch) are generated. \n",
    "- **complexity**: the shape of output arrays produced by each layers of Neural Network models in Generator and Discriminator. The exception are the last layers in both NN models that produce one-dimensional array in case of Discriminator and m-dimensional array for Generator with m equal to the number of rows in train data. This parameter defines the complexity of Neural Network models.\n",
    "- **mode**: WGAN algorithm can be performed in a standard way (‘wgan’) or using a gradient penalty clipping (‘wgan-gp’) introduced by Gulrajani et al. (2017).\n",
    "- **gradient_coeff**: a gradient penalty coefficient that was experimentally tested by Gulrajani et al. (2017) on a variety of NN architectures and datasets. The default value is 10. According to Frigerio L. smaller values of a gradient penalty coefficient increase a speed of training for small datasets but can be compensated by increasing the number of disc_iters instead.\n",
    "- **norm_bound**: L2 norm that is used to clip a gradient of Discriminator. For more details see Abadi et al. (2016).\n",
    "- **learning_rate**: learning rate of the differential private Stochastic Gradient Descent optimizer (dp-SGD) for Discriminator. The default value is 0.001. The learning rate of the Adam optimizer for Generator is not a parameter (but can be implemented if necessary) and by default equal to 0.0001.\n",
    "- **noise**: a noise level that corresponds to a variance of the gaussian noise added to the gradient (dp-SGD) of Discriminator optimizer to achieve differential privacy. This parameter has a direct impact on privacy budget: the higher a noise level, the smaller value of epsilon.\n",
    "- **gen_iters**: the number of Generator iterations to produce one batch of fake samples. Critic_gen in Lorenzo’s version.\n",
    "- **disc_iters**: the number of Discriminator iterations performed by one iteration of Generator and using only one batch of samples.\n",
    "- **target_delta**: differential privacy parameter that defines the probability at which a certain level of privacy (epsilon) is correct. At each generator run the algorithm returns epsilon value. If target_delta is None, it is calculated as following:  1/(number of rows in train data ). The user-defined target_epsilon can be easily implemented if necessary.\n",
    "- **stop_criterion**: the differential privacy budget (epsilon, by default 32). In a differential private mode of WGAN the training is stopped as far as the epsilon value of generated data is higher than a given stopping criterion. If the stopping criterion is assigned to +Inf or WGAN is trained in a public mode, the algorithm will stop after 1000 iterations (epochs). \n",
    "- **interval**: a period with which the WGAN outputs and accuracies of Classifiers (scores and plots) are saved.\n",
    "\n",
    "Use print(my_params.__dict__) to see the default values: \n",
    "\n",
    "```my_params = WGAN_params(noise = 3.0, target_delta = 0.00001)\n",
    "print(my_params.__dict__)```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class WGAN_params(object):\n",
    "    def __init__(self, batch_size = 256, complexity = 512, mode = 'wgan-gp',\n",
    "                 gradient_coeff = 10, norm_bound = 0.2, learning_rate = 1e-3, noise = 7.0,\n",
    "                 gen_iters = 1, disc_iters = 5, target_delta = None, stop_criterion = 32, interval = 20):\n",
    "        self.batch_size = batch_size\n",
    "        self.complexity = complexity\n",
    "        self.mode = mode\n",
    "        self.gradient_coeff = gradient_coeff\n",
    "        self.norm_bound = norm_bound\n",
    "        self.learning_rate = learning_rate\n",
    "        self.noise = noise\n",
    "        self.gen_iters = gen_iters\n",
    "        self.disc_iters = disc_iters\n",
    "        self.target_delta = target_delta\n",
    "        self.stop_criterion = stop_criterion\n",
    "        self.interval = interval\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'batch_size': 256, 'complexity': 512, 'mode': 'wgan-gp', 'gradient_coeff': 10, 'norm_bound': 0.2, 'learning_rate': 0.001, 'noise': 7.0, 'gen_iters': 1, 'disc_iters': 5, 'target_delta': None, 'stop_criterion': 32, 'interval': 2}\n"
     ]
    }
   ],
   "source": [
    "# Set WGAN hyperparameters\n",
    "my_params = WGAN_params(interval = 2)\n",
    "print(my_params.__dict__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "   \n",
    "# Save Cramer_V statistic for association between the same categorical features in generated (fake) vs. input (real) data\n",
    "# A score for each epoch\n",
    "\n",
    "def save_CramerV(list, epoch):\n",
    "    path = './correlation'\n",
    "    if os.path.isdir (path) == False:\n",
    "        os.makedirs(path)\n",
    "    else:\n",
    "        pass\n",
    "    with open(path + \"/CramerV_\"+ str(epoch)+'.txt','w') as thefile:\n",
    "        for item in list:\n",
    "            thefile.write(\"%s\\n\" % item)\n",
    "            \n",
    "# Save the matrix of correlations between different features\n",
    "# Two correlation matrices side by side are returned, i.e. for real and generated data\n",
    "# A matrix for each epoch\n",
    "# WARNING: data should not contain missing values\n",
    "\n",
    "def saveCorrelationMatrix(real, fake, epoch):\n",
    "    path = './correlation'\n",
    "    if os.path.isdir (path) == False:\n",
    "        os.makedirs(path)\n",
    "    else:\n",
    "        pass\n",
    "    real = pd.get_dummies(real).corr(method = 'pearson')\n",
    "    fake = pd.get_dummies(fake).corr(method = 'pearson')\n",
    "    fig, axs = plt.subplots(figsize = (15,7), ncols=2)\n",
    "    sns.heatmap(real, yticklabels = False, ax = axs[0])\n",
    "    axs[0].set_title('Real data')\n",
    "    sns.heatmap(fake, yticklabels = False, ax = axs[1])\n",
    "    axs[1].set_title('Generated data')\n",
    "    fig.savefig(path +\"/CorrelationMatrix_\"+ str(epoch) +\".png\")\n",
    "    plt.close(fig)\n",
    "\n",
    "def getOccurencies(discManager,data):\n",
    "    lengths = discManager.getLabelsLength()\n",
    "    totIndex = 0\n",
    "    columnsValues = discManager.OHdataset.columns.tolist()\n",
    "    Freq = []\n",
    "    Name = []\n",
    "    for index, length in enumerate(lengths):\n",
    "        Freq_per_Category = []\n",
    "        Name_Category = []\n",
    "        if(length > 1):\n",
    "            samplesDisc = np.argmax(data[:, totIndex:totIndex + length], axis=1)\n",
    "            for i in range(length):\n",
    "                Freq_per_Category.append((samplesDisc == i).sum())\n",
    "                Name_Category.append(columnsValues[i + totIndex])\n",
    "        Freq += Freq_per_Category\n",
    "        Name += Name_Category\n",
    "        totIndex += length\n",
    "    FreqData = pd.concat([pd.Series(Name), pd.Series(Freq)], axis = 1)\n",
    "    FreqData.columns = ['Category','Occurrency']\n",
    "    return FreqData"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A function ‘train_WGAN’ requires only two arguments: preprocessed data and a dictionary with hyperparameters of WGAN that are set at the previous step. The algorithm will create automatically a directory (using a given path) where all outputs with a given interval (by default every 20 iterations) will be stored.\n",
    "\n",
    "```train_WGAN(pat, my_params, './dp-WGAN-discrete_patient_occurencies')```\n",
    "\n",
    "\n",
    "![alt text](./figs/image003.png \"Title\")\n",
    "\n",
    "#### Accuracy of AutoML classifier for labeled data\n",
    "\n",
    "If data labelled, AutoML Classifier (H2O.ai) is used to evaluate the quality of differential private data generated by WGAN. There is no implementation for unlabeled data yet. The quality of differential private data generated by WGAN is evaluated following Shmelkov et al. (2018): An important characteristic of GAN model is that generated data should not only be realistic, but also recognizable as coming from a given class. An optimal GAN that perfectly captures the target distribution can generate a new set of data (x_fake), which are indistinguishable from the original real training set (x_train). Assuming both these sets have the same size, a classiﬁer trained on either of them should produce roughly the same validation accuracy.\n",
    "A figure is based on Shmelkov et al. (2018):\n",
    "\n",
    "![alt text](./figs/image006.jpg \"Title\")\n",
    "\n",
    "- **acc_train** is the accuracy of a classiﬁer trained on fake data (x_fake) and tested on a real test data (x_test). When a GAN is not perfect, acc_train will be lower than acc_valid. When acc_train is close to validation accuracy, it means that generated fake data are high quality and as diverse as the training set. acc_train is a measure of the diversity of a GAN, similar to recall, measuring the fraction of real data covered by a GAN.\n",
    "- **acc_test** is the accuracy of a classiﬁer trained on the original train data but tested on fake (generated) data. Ideally, acc_test should be close to the validation accuracy. If it signiﬁcantly higher, it means that the GAN overﬁts, and simply memorizes the training set. On the contrary, if it is signiﬁcantly lower, the GAN does not capture the target distribution well and the data quality is poor. Note that this measure does not capture the diversity of samples. acc_test accuracy is related to the precision score, quantifying how close generated data are to real data.\n",
    "- **acc_valid** is the typical validation accuracy of the classiﬁer trained on real train data (x_train) and validated on real test data (x_test).\n",
    "\n",
    "References:\n",
    "1.\tMartin Abadi, Andy Chu, Ian Goodfellow, H. Brendan McMahan, Ilya Mironov, Kunal Talwar, and Li Zhang. 2016. Deep Learning with Differential Privacy. In Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security (CCS '16). ACM, New York, NY, USA, 308-318. DOI: https://doi.org/10.1145/2976749.2978318.\n",
    "2.\tIshaan Gulrajani, Faruk Ahmed, Martin Arjovsky, Vincent Dumoulin, and Aaron Courville. 2017. Improved training of wasserstein GANs. In Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS'17), Ulrike von Luxburg, Isabelle Guyon, Samy Bengio, Hanna Wallach, and Rob Fergus (Eds.). Curran Associates Inc., USA, 5769-5779.\n",
    "3.\tKonstantin Shmelkov, Cordelia Schmid, and Karteek Alahari. 2018. How good is my GAN? In Proceedings of European Conference on Computer Vision."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# input_data is an output of prepare_data, params is paramters for WGAN training, \n",
    "# path is a string to distinguish different datasets (will be used to create a directiry name)\n",
    "\n",
    "def train_WGAN(input_data, params, path):\n",
    "    \n",
    "    # To show the execution time of WGAN\n",
    "    start_time = time.time()\n",
    "    \n",
    "    # Define number of the training iteration \n",
    "    epoch = 1000\n",
    "    \n",
    "    # Define training data\n",
    "    data = input_data.train_data\n",
    "    # One hot encoder of the input dataset\n",
    "    discManager = discretedata_manager(data)\n",
    "    \n",
    "    # Create a directory to store results\n",
    "    path = path\n",
    "    if os.path.isdir(path) == False:\n",
    "        os.makedirs(path)  \n",
    "    else:\n",
    "        pass\n",
    "    print('Outputs will be saved to:'+ str(path))\n",
    "    \n",
    "    # Define target delta if it is not given\n",
    "    if params.target_delta is None:\n",
    "        target_delta = 1/data.shape[0]\n",
    "    else: target_delta = params.target_delta\n",
    "    \n",
    "    # Build the improved WGAN architecture [Gulrajani et al. 2017]\n",
    "    diminput = discManager.getLabelLengthTot()\n",
    "    real_data = tf.placeholder(tf.float32, shape = [None, diminput])\n",
    "    fake_data = Generator(params.batch_size, real_data, discManager, diminput, params.complexity) \n",
    "    disc_real = Discriminator(real_data, diminput, params.complexity)\n",
    "    disc_fake = Discriminator(fake_data, diminput, params.complexity)\n",
    "    \n",
    "    # WGAN loss functions for Discriminator and Generator\n",
    "    disc_cost = tf.reduce_mean(disc_fake) - tf.reduce_mean(disc_real) \n",
    "    gen_cost = -tf.reduce_mean(disc_fake)\n",
    "    \n",
    "    # WGAN gradient penalty\n",
    "    if params.mode == 'wgan-gp':\n",
    "        alpha = tf.random_uniform(shape = [params.batch_size,1], minval=0.,maxval=1.) \n",
    "        interpolates = alpha*real_data + ((1-alpha)*fake_data)\n",
    "        disc_interpolates = Discriminator(interpolates, diminput, params.complexity)\n",
    "        gradients = tf.gradients(disc_interpolates, [interpolates])[0]\n",
    "        slopes = tf.sqrt(tf.reduce_sum(tf.square(gradients), reduction_indices=[1]))\n",
    "        gradient_penalty = tf.reduce_mean((slopes-1)**2)\n",
    "        disc_cost += params.gradient_coeff*gradient_penalty\n",
    "\n",
    "    disc_params = lib.params_with_name('Discriminator')\n",
    "    gen_params = lib.params_with_name('Generator')\n",
    "\n",
    "    # Hyperparameters of the differential privacy for WGAN\n",
    "    if params.mode == 'wgan-gp':\n",
    "        eps = tf.placeholder(tf.float32)\n",
    "        delta = tf.placeholder(tf.float32)\n",
    "        \n",
    "        priv_accountant = accountant.GaussianMomentsAccountant(len(input_data.train_data))\n",
    "    \n",
    "        clip = True\n",
    "        iteration_decay = tf.placeholder(tf.float32)\n",
    "        l2norm_bound = params.norm_bound  \n",
    "        batches_per_lot = 1\n",
    "        gaussian_sanitizer = sanitizer.AmortizedGaussianSanitizer(priv_accountant, \n",
    "                                                                  [tf.multiply(l2norm_bound,iteration_decay), clip], \n",
    "                                                                  disc_params)\n",
    "        iterationW = tf.placeholder(tf.float32)\n",
    "        tf.summary.scalar(\"iter\", iterationW)\n",
    "    \n",
    "        # Optimizer of the Discriminator (dp-SGD) using differential privacy \n",
    "        disc_train_op, sanitized_grads, clipped_grads, not_sanitized_grads, bound, a  = dp_optimizer.DPGradientDescentOptimizer(\n",
    "            params.learning_rate, \n",
    "            [eps, delta], \n",
    "            sanitizer = gaussian_sanitizer, \n",
    "            sigma = tf.multiply(params.noise,iteration_decay),\n",
    "            batches_per_lot = batches_per_lot).minimize(disc_cost, iteration = iterationW, var_list = disc_params)\n",
    "    \n",
    "#         # Optimizer of the Discriminatort (SGD) without differential privacy\n",
    "#         disc_train_opSGD = tf.train.GradientDescentOptimizer(params.learning_rate).minimize(disc_cost,var_list=disc_params)\n",
    "        \n",
    "#         # Optimizer of the plain GAN with ADAM\n",
    "#         disc_train_opPublic = tf.train.AdamOptimizer(params.learning_rate, beta1=0.5, beta2=0.9).minimize(disc_cost,\n",
    "#                                                                                                           var_list=disc_params)\n",
    "    \n",
    "        # Optimizer of the Generator (ADAM)\n",
    "        if len(gen_params) > 0:\n",
    "            gen_train_op = tf.train.AdamOptimizer(learning_rate=1e-4, beta1 = 0.5, beta2 = 0.9).minimize(gen_cost,\n",
    "                                                                                                         var_list = gen_params)\n",
    "        else:\n",
    "            gen_train_op = tf.no_op()\n",
    "            \n",
    "    # Print parameters of the generator and the discriminator\n",
    "    print (\"Generator params:\")\n",
    "    for var in lib.params_with_name('Generator'):\n",
    "        print (\"\\t{}\\t{}\".format(var.name, var.get_shape()))\n",
    "    print (\"Discriminator params:\")\n",
    "    for var in lib.params_with_name('Discriminator'):\n",
    "        print (\"\\t{}\\t{}\".format(var.name, var.get_shape()))\n",
    "    \n",
    "    # Generate new (fake) data\n",
    "    # Calculate occurrencies of categoies for each feature in a generated data\n",
    "    # Compute the Wasserstein distance on the marginal distibution of the features\n",
    "    \n",
    "    def generate_samples(true_dist, discManager, epoch):\n",
    "        samples = np.concatenate([session.run(fake_data) for _ in range(int(true_dist.shape[0] / int(fake_data.shape[0])))])\n",
    "        discreteSamples, discreteTextDataset = discManager.convertDiscreteMatrix(samples)\n",
    "        FreqTable = getOccurencies(discManager, discreteSamples)\n",
    "        was = 0\n",
    "        for i in range(samples.shape[1]):\n",
    "            was += wasserstein_distance(samples[:, i], true_dist[:, i])\n",
    "        was /= samples.shape[1]\n",
    "        print(\"Wasserstein distance: \" + str(round(was, 4)))\n",
    "        return discreteTextDataset, FreqTable, was\n",
    "\n",
    "    # Compute and display privacy values (delta, epsilon) achieved by dp-WGAN\n",
    "    def plotPrivacyMeasures(session):\n",
    "        spent_deltas_eps = priv_accountant.get_privacy_spent(session, target_deltas = [target_delta])\n",
    "        for (spent_eps, spent_delta) in spent_deltas_eps:\n",
    "            print(\"epsilon = \" + str(spent_eps) + \", delta = \" + str(spent_delta))\n",
    "            return spent_eps\n",
    "    # Save the plot of the Wasserstein distances wrt the time\n",
    "    # A single plot for each epoch\n",
    "    # Progressing over time (number of epochs on a given plot is constantly updating during the training)\n",
    "\n",
    "    def save_plot_was(was):\n",
    "        dirname = os.path.join(path,'plots').replace(\"\\\\\",\"/\")\n",
    "        if os.path.isdir (dirname) == False:\n",
    "            os.makedirs(dirname)\n",
    "        else:\n",
    "            pass\n",
    "        X = list(range(0, len(was)*params.interval, params.interval))\n",
    "        fig, ax = plt.subplots(nrows=1, ncols=1)\n",
    "        ax.plot(X, was)\n",
    "        ax.set_xlabel(\"Number of WGAN iterations\")\n",
    "        ax.set_ylabel(\"Wasserstein distance\")\n",
    "        fig.savefig(os.path.join(dirname,\"wasserstein_plot.png\").replace(\"\\\\\",\"/\"))\n",
    "        plt.close(fig)\n",
    "        \n",
    "    # Save the plot to compare the accuracies of AutoML Classifiers \n",
    "    # Generated (fake) data is used to train Classifier\n",
    "    # Predicted values are compared to real data: training data that where used to train WGAN and test data that were left apart\n",
    "    # A single plot for each epoch\n",
    "    # Progressing over time (number of epochs on a given plot is constantly updating during the training)\n",
    "\n",
    "    def save_plot_accuracy(acc_test, acc_train, acc_valid):\n",
    "        dirname = os.path.join(path,'plots').replace(\"\\\\\",\"/\")\n",
    "        if os.path.isdir(dirname) == False:\n",
    "            os.makedirs(dirname)\n",
    "        else:\n",
    "            pass\n",
    "        X = list(range(0, len(acc_test)*params.interval, params.interval))\n",
    "        fig, ax = plt.subplots(nrows = 1, ncols = 1 )\n",
    "        ax.plot(X, acc_test, label = \"Test\")\n",
    "        ax.plot(X, acc_train, label = \"Train\")\n",
    "        ax.plot(X, acc_valid, label = \"Validation\")\n",
    "        ax.set_xlabel(\"Number of WGAN iterations\")\n",
    "        ax.set_ylabel(\"Accuracy\")\n",
    "        ax.legend(loc = 'lower right')\n",
    "        fig.savefig(os.path.join(dirname, \"acc_plot.png\").replace(\"\\\\\",\"/\"))\n",
    "        plt.close(fig)\n",
    "\n",
    "    # Save accuracy's scores of Random Forest Classifier in a text file\n",
    "    # All scores for different epochs are stored to a single file\n",
    "\n",
    "    def save_accuracy(list, name_file):\n",
    "        dirname = os.path.join(path,'scores').replace(\"\\\\\",\"/\")\n",
    "        if os.path.isdir (dirname) == False:\n",
    "            os.makedirs(dirname)\n",
    "        else:\n",
    "            pass\n",
    "        with open(os.path.join(dirname, str(name_file)+'.txt').replace(\"\\\\\",\"/\"),'w') as thefile:\n",
    "            for item in list:\n",
    "                thefile.write(\"%s\\n\" % item)\n",
    "            \n",
    "#     # Save WGAN model\n",
    "#     saver = tf.train.Saver()\n",
    "    \n",
    "    # Train the WGAN model\n",
    "    with tf.Session() as session:\n",
    "        session.run(tf.global_variables_initializer())\n",
    "        # Get the one-hot encoding of the input data\n",
    "        data = discManager.OHdataset.values\n",
    "        # Get category's occurrencies in original dataset\n",
    "        OriginFreqData = getOccurencies(discManager, data)  \n",
    "        OriginFreqData.to_csv(os.path.join(path, \"OriginalFreq.csv\").replace(\"\\\\\",\"/\"), index = False)\n",
    "        # Shuffle the input data\n",
    "        np.random.shuffle(data)                     \n",
    "        accuracies_test = []\n",
    "        accuracies_train = []\n",
    "        accuracies_valid = []\n",
    "        dp_epsilon = []\n",
    "        wasTot = []\n",
    "        best_was = np.inf\n",
    "        _old_gen_cost = 0\n",
    "        _gen_cost = 0\n",
    "        count_change = 0\n",
    "        stop_criterion = 0\n",
    "        # Do n epochs \n",
    "        for iteration in range(epoch):\n",
    "            # Early Stopping of WGAN training when a privacy budget is exhausted:\n",
    "            if stop_criterion < params.stop_criterion:\n",
    "                print(\"\\nEpoch \"+str(iteration+1))\n",
    "                # Train the Critic\n",
    "                for idx in range(int(data.shape[0] / (params.batch_size * params.disc_iters))):\n",
    "                    # print(\"Batch \"+str(idx+1))\n",
    "                    # Train the Generator\n",
    "                    if iteration > 0:\n",
    "                        for _ in range(params.gen_iters):\n",
    "                            _gen_cost, _ = session.run([gen_cost, gen_train_op], feed_dict={iterationW: (float(iteration + 1))})\n",
    "                    for i in range(params.disc_iters):\n",
    "                        # print(\"Discriminator iteration \"+str(i+1))\n",
    "\n",
    "                        # Clipping decay\n",
    "                        curr_iter =  (iteration % 5)\n",
    "                        iter_decay = 1 - int(iteration / 10)*0.1\n",
    "                        if (iter_decay <= 0.40):\n",
    "                            iter_decay = 0.35\n",
    "\n",
    "                        # Train the Discriminator\n",
    "                        indeces = np.random.randint(0, data.shape[0] - 1, params.batch_size)\n",
    "                        _disc_cost, _ = session.run([disc_cost, disc_train_op],\n",
    "                                                    feed_dict = {real_data: data[indeces, :], iterationW: (float(curr_iter)),\n",
    "                                                    iteration_decay: iter_decay})  \n",
    "                # Shuffle the data after each training to change the batches\n",
    "                np.random.shuffle(data)\n",
    "                print(\"Discriminator loss: \" + str(round(_disc_cost, 4)))\n",
    "                # Additional dp-GAN imrpovement to save privacy every 20 iterations\n",
    "                if iteration % params.interval == 0:\n",
    "                    if (abs(_gen_cost) < abs(_old_gen_cost)):\n",
    "                        count_change += 1\n",
    "                        if (count_change > 1 and params.disc_iters > 1):\n",
    "                            params.disc_iters -= 1\n",
    "                            count_change = 0\n",
    "                        print(\"change\")\n",
    "                    _old_gen_cost = _gen_cost\n",
    "                    # Compute, display and save differential privacy (epsilon, delta) parameters\n",
    "                    stop_criterion = plotPrivacyMeasures(session)\n",
    "                    dp_epsilon.append(stop_criterion)\n",
    "                    save_accuracy(dp_epsilon, 'epsilon')\n",
    "                    # Generate a new (fake) data\n",
    "                    # Get category's occurrencies in generated dataset\n",
    "                    # Compute and dipslay Wasserstein distance\n",
    "                    output, FakeFreqData, was = generate_samples(data, discManager, iteration)\n",
    "                    # Save generated data to a text file\n",
    "                    anonymized_data = pd.DataFrame(np.array(output).T, columns=discManager.variables)\n",
    "                    # Convert back missing values to NA for generated data\n",
    "                    output_na = anonymized_data\n",
    "                    for col, name in enumerate(output_na.columns):\n",
    "                        output_na.replace(str(\"missing \"+ output_na.columns[col]), np.nan, inplace=True)\n",
    "                    output_na.to_csv(os.path.join(path, \"GeneratedData_epoch_\"+str(iteration+1)+\".csv\").replace(\"\\\\\",\"/\"), \n",
    "                                     index = False)\n",
    "                    # Save category's occurencies for each feature in a generated dataset\n",
    "                    FakeFreqData.to_csv(os.path.join(path, \"GeneratedFreq_epoch_\"+ str(iteration+1)+\".csv\").replace(\"\\\\\",\"/\"), \n",
    "                                        index = False)\n",
    "                    # Save Wasserstein distance, i.e. the plot and values\n",
    "                    wasTot.append(was)\n",
    "                    save_accuracy(wasTot, 'wasserstein_distance')\n",
    "                    save_plot_was(wasTot)\n",
    "\n",
    "                    # If data labelled, calculate the accuracy scores of Classifier using H2O AutoML\n",
    "                    # Train and test accuracy of WGAN are calculated following Shmelkov et al. (2018) How good is my GAN?\n",
    "                    if input_data.target is not None:\n",
    "                        print('\\nData is labelled \\nCalculating accuracy scores for AutoML Classifiers ......')\n",
    "                        y = input_data.target\n",
    "                        # Prepare fake (generated) data\n",
    "                        h2o_fake = h2o.H2OFrame(anonymized_data)\n",
    "                        h2o_fake[y] = h2o_fake[y].asfactor()\n",
    "                        x_fake = h2o_fake.columns\n",
    "                        x_fake.remove(y)\n",
    "                        # Prepaer real train data\n",
    "                        h2o_train = h2o.H2OFrame(input_data.train_data)\n",
    "                        h2o_train[y] = h2o_train[y].asfactor()\n",
    "                        x_train = h2o_train.columns\n",
    "                        x_train.remove(y)\n",
    "                        # Prepare real test data\n",
    "                        h2o_test = h2o.H2OFrame(input_data.test_data)\n",
    "                        h2o_test[y] = h2o_test[y].asfactor()\n",
    "                        # Train WGAN-train-classifier\n",
    "                        aml_train = H2OAutoML(max_models = 10, seed = 1)\n",
    "                        aml_train.train(x = x_fake, y = y, training_frame = h2o_fake)\n",
    "                        # Train WGAN-test-classifier\n",
    "                        aml_test = H2OAutoML(max_models = 10, seed = 1)\n",
    "                        aml_test.train(x = x_train, y = y, training_frame = h2o_train)\n",
    "                        # Calculate a typical validation accuracy\n",
    "                        preds_valid = aml_test.leader.predict(h2o_test)                    \n",
    "                        acc_valid = accuracy_score(h2o_test[y].as_data_frame().values, \n",
    "                                                   preds_valid['predict'].as_data_frame().values)\n",
    "                        # Calculate WGAN-train accuracy\n",
    "                        preds_train = aml_train.leader.predict(h2o_test)                    \n",
    "                        acc_train = accuracy_score(h2o_test[y].as_data_frame().values, \n",
    "                                                   preds_train['predict'].as_data_frame().values)\n",
    "                        # Calculate WGAN-test accuracy\n",
    "                        preds_test = aml_test.leader.predict(h2o_fake)\n",
    "                        acc_test = accuracy_score(h2o_fake[y].as_data_frame().values, \n",
    "                                                   preds_test['predict'].as_data_frame().values)\n",
    "                        # Save both AutoMl Classifiers as MOJO models\n",
    "                        aml_test.leader.download_mojo(path = path +\"/AutoML/test_model_\"+str(iteration+1)+\".zip\")\n",
    "                        aml_train.leader.download_mojo(path = path +\"/AutoML/train_model_\"+str(iteration+1)+\".zip\")\n",
    "                        # Print the results\n",
    "                        print('Validation accuracy: '+ str(acc_valid))\n",
    "                        print('Train accuracy: '+ str(acc_train))\n",
    "                        print('Test accuracy: '+ str(acc_test))\n",
    "                        accuracies_test.append(acc_test)\n",
    "                        accuracies_train.append(acc_train)\n",
    "                        accuracies_valid.append(acc_valid)\n",
    "                        # Store the results of the classification analysis based on generated data\n",
    "                        save_accuracy(accuracies_test,'accuracy_test')\n",
    "                        save_accuracy(accuracies_train, 'accuracy_train')\n",
    "                        save_accuracy(accuracies_valid, 'accuracy_valid')\n",
    "                        save_plot_accuracy(accuracies_test, accuracies_train, accuracies_valid)\n",
    "    # Print info why training is stopped\n",
    "    if stop_criterion > params.stop_criterion:\n",
    "        print(\"\\nPrivacy budget is exhausted ...\")\n",
    "        print(\"WGAN training is stopped ...\")\n",
    "    else: print(\"WGAN training is stopped after\" +str(epoch) +\" iterations ...\")\n",
    "    # Return the execution time of WGAN\n",
    "    execution_time = time.time() - start_time\n",
    "    print(\"\\nThe execution time: \"+str(round(execution_time/60,))+\" min \"+str(round(execution_time%60,2))+\" sec\")\n",
    "    return stop_criterion\n",
    "                    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Outputs will be saved to:./adult_test_stop_criterion\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "W0816 14:44:06.971859 35388 deprecation_wrapper.py:119] From C:\\Users\\i061767\\GitHub\\testbed\\models.py:34: The name tf.random_normal is deprecated. Please use tf.random.normal instead.\n",
      "\n",
      "W0816 14:44:07.348065 35388 deprecation.py:323] From C:\\Users\\i061767\\GitHub\\testbed\\differential_privacy\\dp_sgd\\dp_optimizer\\sanitizers\\basic.py:34: div (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Deprecated in favor of operator or tf.math.divide.\n",
      "W0816 14:44:07.364859 35388 deprecation_wrapper.py:119] From C:\\Users\\i061767\\GitHub\\testbed\\differential_privacy\\dp_sgd\\dp_optimizer\\sanitizers\\basic.py:36: The name tf.rsqrt is deprecated. Please use tf.math.rsqrt instead.\n",
      "\n",
      "W0816 14:44:07.377858 35388 deprecation_wrapper.py:119] From C:\\Users\\i061767\\GitHub\\testbed\\differential_privacy\\dp_sgd\\dp_optimizer\\sanitizers\\basic.py:38: The name tf.diag is deprecated. Please use tf.linalg.tensor_diag instead.\n",
      "\n",
      "W0816 14:44:07.393857 35388 deprecation_wrapper.py:119] From C:\\Users\\i061767\\GitHub\\testbed\\differential_privacy\\privacy_accountant\\tf\\accountant.py:402: The name tf.log is deprecated. Please use tf.math.log instead.\n",
      "\n",
      "W0816 14:44:07.440852 35388 deprecation_wrapper.py:119] From C:\\Users\\i061767\\GitHub\\testbed\\differential_privacy\\privacy_accountant\\tf\\accountant.py:238: The name tf.assign_add is deprecated. Please use tf.compat.v1.assign_add instead.\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generator params:\n",
      "\tGenerator.1.Linear/Generator.1.Linear.W:0\t(104, 512)\n",
      "\tGenerator.1.Linear/Generator.1.Linear.b:0\t(512,)\n",
      "\tGenerator.2.Linear/Generator.2.Linear.W:0\t(512, 512)\n",
      "\tGenerator.2.Linear/Generator.2.Linear.b:0\t(512,)\n",
      "\tGenerator.3.Linear/Generator.3.Linear.W:0\t(512, 512)\n",
      "\tGenerator.3.Linear/Generator.3.Linear.b:0\t(512,)\n",
      "\tGenerator.4/Generator.4.W:0\t(512, 104)\n",
      "\tGenerator.4/Generator.4.b:0\t(104,)\n",
      "Discriminator params:\n",
      "\tDiscriminator.1.Linear/Discriminator.1.Linear.W:0\t(104, 512)\n",
      "\tDiscriminator.1.Linear/Discriminator.1.Linear.b:0\t(512,)\n",
      "\tDiscriminator.2.Linear/Discriminator.2.Linear.W:0\t(512, 512)\n",
      "\tDiscriminator.2.Linear/Discriminator.2.Linear.b:0\t(512,)\n",
      "\tDiscriminator.3.Linear/Discriminator.3.Linear.W:0\t(512, 512)\n",
      "\tDiscriminator.3.Linear/Discriminator.3.Linear.b:0\t(512,)\n",
      "\tDiscriminator.4/Discriminator.4.W:0\t(512, 1)\n",
      "\tDiscriminator.4/Discriminator.4.b:0\t(1,)\n",
      "\n",
      "Epoch 1\n",
      "Discriminator loss: -0.0777\n",
      "epsilon = 0.44882481316968703, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.1115\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7832302936630603\n",
      "Train accuracy: 0.5030911901081917\n",
      "Test accuracy: 0.49799923780487804\n",
      "\n",
      "Epoch 2\n",
      "Discriminator loss: -0.1587\n",
      "\n",
      "Epoch 3\n",
      "Discriminator loss: -0.3412\n",
      "epsilon = 0.7450729313652448, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0987\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7826506955177743\n",
      "Train accuracy: 0.5015455950540958\n",
      "Test accuracy: 0.3468940548780488\n",
      "\n",
      "Epoch 4\n",
      "Discriminator loss: -0.3759\n",
      "\n",
      "Epoch 5\n",
      "Discriminator loss: -0.284\n",
      "epsilon = 0.965265575751333, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0757\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7838098918083463\n",
      "Train accuracy: 0.4992272024729521\n",
      "Test accuracy: 0.5370617378048781\n",
      "\n",
      "Epoch 6\n",
      "Discriminator loss: -0.4477\n",
      "\n",
      "Epoch 7\n",
      "Discriminator loss: -0.4452\n",
      "change\n",
      "epsilon = 1.1459249801624398, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0771\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Columbia, Germany, Greece, India, Ireland, Thailand, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785935085007728\n",
      "Train accuracy: 0.5057959814528593\n",
      "Test accuracy: 0.4445503048780488\n",
      "\n",
      "Epoch 8\n",
      "Discriminator loss: -0.3421\n",
      "\n",
      "Epoch 9\n",
      "Discriminator loss: -0.3912\n",
      "change\n",
      "epsilon = 1.303304892171412, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0798\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "drf prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Columbia, France, Greece, Honduras, Italy, Laos, Philippines, Thailand]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.4186630602782071\n",
      "Test accuracy: 0.9571265243902439\n",
      "\n",
      "Epoch 10\n",
      "Discriminator loss: -0.5416\n",
      "\n",
      "Epoch 11\n",
      "Discriminator loss: -0.7678\n",
      "change\n",
      "epsilon = 1.4606862108378964, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.1017\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "deeplearning prediction progress: |███████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [England, Germany, Ireland, Italy, Japan, Laos, Portugal, South, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7828438948995363\n",
      "Train accuracy: 0.4748840803709428\n",
      "Test accuracy: 0.9109184451219512\n",
      "\n",
      "Epoch 12\n",
      "Discriminator loss: -0.8238\n",
      "\n",
      "Epoch 13\n",
      "Discriminator loss: -0.6645\n",
      "epsilon = 1.6191765629261063, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0825\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "deeplearning prediction progress: |███████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, China, Dominican-Republic, Ecuador, Greece, Honduras, Hungary, India, Iran, Italy, Japan, Outlying-US(Guam-USVI-etc), Philippines, Poland, Portugal, Puerto-Rico, Thailand, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7832302936630603\n",
      "Train accuracy: 0.6207496136012365\n",
      "Test accuracy: 0.06916920731707317\n",
      "\n",
      "Epoch 14\n",
      "Discriminator loss: -0.6958\n",
      "\n",
      "Epoch 15\n",
      "Discriminator loss: -0.4152\n",
      "epsilon = 1.7646960601629493, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0743\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Canada, China, Columbia, Cuba, Dominican-Republic, El-Salvador, England, France, Germany, Greece, Guatemala, Honduras, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Philippines, Poland, Portugal, Puerto-Rico, Yugoslavia]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'relationship' has levels not trained on: [Other-relative]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785355486862442\n",
      "Train accuracy: 0.499806800618238\n",
      "Test accuracy: 0.6782583841463414\n",
      "\n",
      "Epoch 16\n",
      "Discriminator loss: -0.596\n",
      "\n",
      "Epoch 17\n",
      "Discriminator loss: -1.0654\n",
      "epsilon = 1.9005553928636538, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0925\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "deeplearning prediction progress: |███████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Canada, China, Columbia, Cuba, England, France, Germany, Greece, Hong, Iran, Ireland, Italy, Japan, Outlying-US(Guam-USVI-etc), Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Yugoslavia]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Tech-support]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7828438948995363\n",
      "Train accuracy: 0.490919629057187\n",
      "Test accuracy: 0.9427400914634146\n",
      "\n",
      "Epoch 18\n",
      "Discriminator loss: -0.9522\n",
      "\n",
      "Epoch 19\n",
      "Discriminator loss: -0.2841\n",
      "epsilon = 2.029902859369843, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0715\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Canada, Columbia, Cuba, Dominican-Republic, Germany, Greece, Honduras, India, Italy, Jamaica, Outlying-US(Guam-USVI-etc), Portugal, Puerto-Rico, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7834234930448223\n",
      "Train accuracy: 0.5\n",
      "Test accuracy: 0.7799161585365854\n",
      "\n",
      "Epoch 20\n",
      "Discriminator loss: -0.0964\n",
      "\n",
      "Epoch 21\n",
      "Discriminator loss: -0.4822\n",
      "change\n",
      "epsilon = 2.164951649413228, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0693\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "deeplearning prediction progress: |███████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Canada, Columbia, Cuba, Dominican-Republic, Germany, Greece, Honduras, Hong, Hungary, India, Ireland, Italy, Japan, Laos, Outlying-US(Guam-USVI-etc), Puerto-Rico, Taiwan, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785548686244204\n",
      "Train accuracy: 0.44068778979907264\n",
      "Test accuracy: 0.06554878048780488\n",
      "\n",
      "Epoch 22\n",
      "Discriminator loss: -0.2654\n",
      "\n",
      "Epoch 23\n",
      "Discriminator loss: -0.4577\n",
      "epsilon = 2.3099278837166635, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.1059\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "drf prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'workclass' has levels not trained on: [Never-worked]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Columbia, Greece, India, Iran, Italy, Laos, Outlying-US(Guam-USVI-etc), Puerto-Rico, South]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.5\n",
      "Test accuracy: 0.016196646341463415\n",
      "\n",
      "Epoch 24\n",
      "Discriminator loss: -1.1896\n",
      "\n",
      "Epoch 25\n",
      "Discriminator loss: -0.4345\n",
      "epsilon = 2.4405298508794826, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0785\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "deeplearning prediction progress: |███████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, China, Columbia, Cuba, Dominican-Republic, Ecuador, England, Germany, Greece, Honduras, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Outlying-US(Guam-USVI-etc), Philippines, Poland, Portugal, Puerto-Rico, South, Vietnam]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7824574961360123\n",
      "Train accuracy: 0.4860896445131376\n",
      "Test accuracy: 0.8835746951219512\n",
      "\n",
      "Epoch 26\n",
      "Discriminator loss: -0.4544\n",
      "\n",
      "Epoch 27\n",
      "Discriminator loss: -0.3609\n",
      "epsilon = 2.571481825018756, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0715\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [?, Cambodia, Canada, China, Columbia, Cuba, Dominican-Republic, Ecuador, France, Germany, Greece, Honduras, Hong, Hungary, India, Ireland, Italy, Japan, Laos, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7836166924265843\n",
      "Train accuracy: 0.47372488408037094\n",
      "Test accuracy: 0.3669969512195122\n",
      "\n",
      "Epoch 28\n",
      "Discriminator loss: -0.3496\n",
      "\n",
      "Epoch 29\n",
      "Discriminator loss: -0.5908\n",
      "change\n",
      "epsilon = 2.6949682933744623, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0714\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [11th, 1st-4th]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, Canada, Columbia, Cuba, Dominican-Republic, England, France, Germany, Greece, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Laos, Outlying-US(Guam-USVI-etc), Philippines, Poland, Portugal, Puerto-Rico, South, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7838098918083463\n",
      "Train accuracy: 0.5\n",
      "Test accuracy: 0.9077743902439024\n",
      "\n",
      "Epoch 30\n",
      "Discriminator loss: -0.3454\n",
      "\n",
      "Epoch 31\n",
      "Discriminator loss: -0.8088\n",
      "epsilon = 2.8295169322698195, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0794\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, Canada, Columbia, Cuba, Dominican-Republic, England, Germany, Greece, Honduras, Hong, India, Ireland, Italy, Japan, Laos, Outlying-US(Guam-USVI-etc), Portugal, Puerto-Rico, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [1st-4th, Prof-school]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.499613601236476\n",
      "Test accuracy: 0.09108231707317073\n",
      "\n",
      "Epoch 32\n",
      "Discriminator loss: -0.4523\n",
      "\n",
      "Epoch 33\n",
      "Discriminator loss: -0.4652\n",
      "epsilon = 2.9823630429659196, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0696\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, China, Columbia, Cuba, Dominican-Republic, Ecuador, El-Salvador, England, Germany, Greece, Hong, Hungary, India, Iran, Ireland, Jamaica, Japan, Laos, Outlying-US(Guam-USVI-etc), Peru, Portugal, Puerto-Rico, South, Taiwan, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [12th, 1st-4th, Prof-school]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7836166924265843\n",
      "Train accuracy: 0.5038639876352395\n",
      "Test accuracy: 0.23256478658536586\n",
      "\n",
      "Epoch 34\n",
      "Discriminator loss: -0.5585\n",
      "\n",
      "Epoch 35\n",
      "Discriminator loss: -0.5448\n",
      "change\n",
      "epsilon = 3.1272154749576493, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.066\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [1st-4th]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, Canada, China, Columbia, Cuba, Dominican-Republic, Ecuador, El-Salvador, England, Germany, Greece, Honduras, Hungary, Ireland, Italy, Jamaica, Japan, Laos, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7834234930448223\n",
      "Train accuracy: 0.5\n",
      "Test accuracy: 0.4699885670731707\n",
      "\n",
      "Epoch 36\n",
      "Discriminator loss: -0.147\n",
      "\n",
      "Epoch 37\n",
      "Discriminator loss: -0.4157\n",
      "epsilon = 3.2641512530492895, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0885\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "deeplearning prediction progress: |███████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Farming-fishing, Handlers-cleaners, Priv-house-serv, Tech-support]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, Canada, China, Columbia, Cuba, Dominican-Republic, Ecuador, El-Salvador, England, France, Germany, Greece, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'marital-status' has levels not trained on: [Married-AF-spouse, Separated]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785548686244204\n",
      "Train accuracy: 0.5539026275115919\n",
      "Test accuracy: 0.31488185975609756\n",
      "\n",
      "Epoch 38\n",
      "Discriminator loss: -0.5303\n",
      "\n",
      "Epoch 39\n",
      "Discriminator loss: -0.8419\n",
      "change\n",
      "epsilon = 3.40108703114093, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.1062\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, China, Columbia, Cuba, Dominican-Republic, El-Salvador, England, Germany, Greece, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Jamaica, Japan, Laos, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [?, Armed-Forces, Farming-fishing, Priv-house-serv, Tech-support]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [11th, 12th, 1st-4th, 7th-8th, 9th, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'race' has levels not trained on: [Asian-Pac-Islander]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7832302936630603\n",
      "Train accuracy: 0.625772797527048\n",
      "Test accuracy: 0.03706173780487805\n",
      "\n",
      "Epoch 40\n",
      "Discriminator loss: -0.6188\n",
      "\n",
      "Epoch 41\n",
      "Discriminator loss: -0.6383\n",
      "epsilon = 3.5636501137156906, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.088\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, China, Cuba, Dominican-Republic, Ecuador, England, Germany, Greece, Honduras, Hungary, India, Iran, Ireland, Jamaica, Japan, Laos, Outlying-US(Guam-USVI-etc), Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Farming-fishing, Tech-support]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [11th, 12th, 1st-4th, 5th-6th, 9th, Assoc-acdm, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'race' has levels not trained on: [Amer-Indian-Eskimo]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7847758887171561\n",
      "Train accuracy: 0.5019319938176198\n",
      "Test accuracy: 0.963891006097561\n",
      "\n",
      "Epoch 42\n",
      "Discriminator loss: -0.551\n",
      "\n",
      "Epoch 43\n",
      "Discriminator loss: -0.0349\n",
      "epsilon = 3.7422876503718614, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0615\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [?, Cambodia, Canada, China, Columbia, Cuba, Dominican-Republic, Ecuador, England, Germany, Greece, Haiti, Honduras, Hungary, India, Ireland, Italy, Jamaica, Japan, Laos, Outlying-US(Guam-USVI-etc), Poland, Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Priv-house-serv, Tech-support]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'race' has levels not trained on: [Amer-Indian-Eskimo, Asian-Pac-Islander]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7828438948995363\n",
      "Train accuracy: 0.7045981452859351\n",
      "Test accuracy: 0.6929306402439024\n",
      "\n",
      "Epoch 44\n",
      "Discriminator loss: -0.2521\n",
      "\n",
      "Epoch 45\n",
      "Discriminator loss: -0.2397\n",
      "epsilon = 3.9032145707580073, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0681\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Farming-fishing, Priv-house-serv, Tech-support, Transport-moving]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [?, Cambodia, Canada, China, Columbia, Cuba, Ecuador, England, Germany, Greece, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.6742658423493045\n",
      "Test accuracy: 0.846608231707317\n",
      "\n",
      "Epoch 46\n",
      "Discriminator loss: -0.1688\n",
      "\n",
      "Epoch 47\n",
      "Discriminator loss: -0.5774\n",
      "change\n",
      "epsilon = 4.064141491144153, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0654\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [1st-4th, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, China, Columbia, Cuba, Dominican-Republic, Ecuador, England, Germany, Greece, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Jamaica, Japan, Laos, Outlying-US(Guam-USVI-etc), Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Farming-fishing, Priv-house-serv, Tech-support]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'marital-status' has levels not trained on: [Married-AF-spouse, Married-spouse-absent, Separated]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7814914992272025\n",
      "Train accuracy: 0.6597758887171561\n",
      "Test accuracy: 0.7052210365853658\n",
      "\n",
      "Epoch 48\n",
      "Discriminator loss: -0.4687\n",
      "\n",
      "Epoch 49\n",
      "Discriminator loss: 0.1192\n",
      "epsilon = 4.229091584539953, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0917\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'workclass' has levels not trained on: [Local-gov, Never-worked]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [11th, 1st-4th, 7th-8th, 9th, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'marital-status' has levels not trained on: [Divorced, Married-AF-spouse, Married-spouse-absent, Separated]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Farming-fishing, Machine-op-inspct, Priv-house-serv, Tech-support]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [?, Cambodia, Canada, China, Columbia, Cuba, Dominican-Republic, Ecuador, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7838098918083463\n",
      "Train accuracy: 0.5525502318392581\n",
      "Test accuracy: 0.8881478658536586\n",
      "\n",
      "Epoch 50\n",
      "Discriminator loss: -0.2676\n",
      "\n",
      "Epoch 51\n",
      "Discriminator loss: -0.3976\n",
      "change\n",
      "epsilon = 4.4321373742114645, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0805\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [11th, 12th, 1st-4th, 5th-6th, 9th, Assoc-acdm, Assoc-voc, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'workclass' has levels not trained on: [Federal-gov, Never-worked]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Craft-repair, Farming-fishing, Handlers-cleaners, Machine-op-inspct, Priv-house-serv, Prof-specialty, Tech-support, Transport-moving]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [?, Cambodia, Canada, China, Columbia, Cuba, Dominican-Republic, Ecuador, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785741885625966\n",
      "Train accuracy: 0.6879829984544049\n",
      "Test accuracy: 0.9154916158536586\n",
      "\n",
      "Epoch 52\n",
      "Discriminator loss: 0.017\n",
      "\n",
      "Epoch 53\n",
      "Discriminator loss: -0.4531\n",
      "epsilon = 4.661647898491367, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0805\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [11th, 12th, 1st-4th, 5th-6th, 9th, Assoc-acdm, Doctorate, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Craft-repair, Handlers-cleaners, Machine-op-inspct, Other-service, Priv-house-serv, Prof-specialty, Protective-serv, Tech-support, Transport-moving]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [?, Cambodia, China, Columbia, Cuba, Dominican-Republic, Ecuador, El-Salvador, England, France, Germany, Greece, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'marital-status' has levels not trained on: [Married-AF-spouse, Married-spouse-absent, Separated, Widowed]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7840030911901081\n",
      "Train accuracy: 0.5778593508500772\n",
      "Test accuracy: 0.2781059451219512\n",
      "\n",
      "Epoch 54\n",
      "Discriminator loss: -0.2871\n",
      "\n",
      "Epoch 55\n",
      "Discriminator loss: 0.0702\n",
      "epsilon = 4.861904726524638, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0635\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Craft-repair, Handlers-cleaners, Machine-op-inspct, Priv-house-serv, Tech-support, Transport-moving]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [11th, 12th, 1st-4th, 5th-6th, Assoc-acdm, Assoc-voc, Masters, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'workclass' has levels not trained on: [Never-worked, Self-emp-not-inc]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, China, Columbia, Cuba, Dominican-Republic, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785355486862442\n",
      "Train accuracy: 0.6570710973724884\n",
      "Test accuracy: 0.5299161585365854\n",
      "\n",
      "Epoch 56\n",
      "Discriminator loss: -0.4173\n",
      "\n",
      "Epoch 57\n",
      "Discriminator loss: -0.5956\n",
      "change\n",
      "epsilon = 5.062161554557909, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0974\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'workclass' has levels not trained on: [?, Never-worked, Self-emp-inc, Self-emp-not-inc]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [11th, 12th, 1st-4th, 5th-6th, Assoc-acdm, Assoc-voc, Doctorate, Masters, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Craft-repair, Priv-house-serv, Tech-support]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'relationship' has levels not trained on: [Other-relative, Wife]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'race' has levels not trained on: [Other]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [?, Cambodia, China, Columbia, Cuba, Dominican-Republic, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Mexico, Nicaragua, Outlying-US(Guam-USVI-etc), Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.5958268933539412\n",
      "Test accuracy: 0.8138338414634146\n",
      "\n",
      "Epoch 58\n",
      "Discriminator loss: -0.0755\n",
      "\n",
      "Epoch 59\n",
      "Discriminator loss: -0.0861\n",
      "epsilon = 5.2624183825911794, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0944\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "drf prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [11th, 12th, 1st-4th, 5th-6th, Assoc-acdm, Assoc-voc, Bachelors, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Craft-repair, Farming-fishing, Other-service, Tech-support]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'relationship' has levels not trained on: [Other-relative, Unmarried, Wife]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [?, Cambodia, China, Columbia, Cuba, Dominican-Republic, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Mexico, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.6945517774343122\n",
      "Test accuracy: 0.5061928353658537\n",
      "\n",
      "Epoch 60\n",
      "Discriminator loss: -0.2062\n",
      "\n",
      "Epoch 61\n",
      "Discriminator loss: -0.3349\n",
      "change\n",
      "epsilon = 5.5771706113881265, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0529\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "deeplearning prediction progress: |███████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [11th, 12th, 1st-4th, 5th-6th, Assoc-acdm, Doctorate, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Craft-repair, Farming-fishing, Other-service, Priv-house-serv, Protective-serv, Tech-support]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'workclass' has levels not trained on: [Never-worked, State-gov]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [?, Cambodia, Canada, China, Columbia, Cuba, Dominican-Republic, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Mexico, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7845826893353941\n",
      "Train accuracy: 0.6821870170015456\n",
      "Test accuracy: 0.8325076219512195\n",
      "\n",
      "Epoch 62\n",
      "Discriminator loss: -0.2763\n",
      "\n",
      "Epoch 63\n",
      "Discriminator loss: -0.2415\n",
      "epsilon = 6.00641824094875, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0563\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'race' has levels not trained on: [Asian-Pac-Islander, Other]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7828438948995363\n",
      "Train accuracy: 0.6883693972179289\n",
      "Test accuracy: 0.7811547256097561\n",
      "\n",
      "Epoch 64\n",
      "Discriminator loss: 0.0004\n",
      "\n",
      "Epoch 65\n",
      "Discriminator loss: -0.4606\n",
      "epsilon = 6.370235182648969, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0604\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785548686244204\n",
      "Train accuracy: 0.741499227202473\n",
      "Test accuracy: 0.8163109756097561\n",
      "\n",
      "Epoch 66\n",
      "Discriminator loss: -0.2046\n",
      "\n",
      "Epoch 67\n",
      "Discriminator loss: -0.2134\n",
      "change\n",
      "epsilon = 6.711022379849122, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0509\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Priv-house-serv, Protective-serv, Tech-support]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [?, Cambodia, China, Columbia, Cuba, Dominican-Republic, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Mexico, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7834234930448223\n",
      "Train accuracy: 0.6526275115919629\n",
      "Test accuracy: 0.8060213414634146\n",
      "\n",
      "Epoch 68\n",
      "Discriminator loss: -0.568\n",
      "\n",
      "Epoch 69\n",
      "Discriminator loss: -0.042\n",
      "epsilon = 7.051809577049274, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0498\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Farming-fishing, Handlers-cleaners, Priv-house-serv, Protective-serv, Tech-support]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [11th, 12th, 1st-4th, 5th-6th, 9th, Assoc-acdm, Assoc-voc, Doctorate, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [?, Cambodia, China, Columbia, Cuba, Dominican-Republic, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.6493431221020093\n",
      "Test accuracy: 0.7286585365853658\n",
      "\n",
      "Epoch 70\n",
      "Discriminator loss: -0.3771\n",
      "\n",
      "Epoch 71\n",
      "Discriminator loss: -0.4116\n",
      "change\n",
      "epsilon = 7.3925967742494265, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0728\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [?, Armed-Forces, Farming-fishing, Handlers-cleaners, Priv-house-serv, Protective-serv, Tech-support, Transport-moving]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'race' has levels not trained on: [Amer-Indian-Eskimo, Black, Other]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7849690880989181\n",
      "Train accuracy: 0.7173493044822257\n",
      "Test accuracy: 0.7019817073170732\n",
      "\n",
      "Epoch 72\n",
      "Discriminator loss: -0.2382\n",
      "\n",
      "Epoch 73\n",
      "Discriminator loss: -0.2329\n",
      "epsilon = 7.733383971449579, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0564\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [11th, 12th, 1st-4th, 5th-6th, 7th-8th, 9th, Assoc-acdm, Assoc-voc, Doctorate, Masters, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [?, Cambodia, China, Columbia, Cuba, Dominican-Republic, Ecuador, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'race' has levels not trained on: [Amer-Indian-Eskimo, Other]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.6317619783616693\n",
      "Test accuracy: 0.8696646341463414\n",
      "\n",
      "Epoch 74\n",
      "Discriminator loss: -0.0788\n",
      "\n",
      "Epoch 75\n",
      "Discriminator loss: -0.3389\n",
      "epsilon = 8.074171168649732, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0577\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [11th, 12th, 1st-4th, 5th-6th, 7th-8th, 9th, Assoc-acdm, Assoc-voc, Doctorate, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Farming-fishing, Handlers-cleaners, Priv-house-serv, Protective-serv, Tech-support, Transport-moving]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7849690880989181\n",
      "Train accuracy: 0.6081916537867079\n",
      "Test accuracy: 0.6436737804878049\n",
      "\n",
      "Epoch 76\n",
      "Discriminator loss: -0.1913\n",
      "\n",
      "Epoch 77\n",
      "Discriminator loss: -0.628\n",
      "change\n",
      "epsilon = 8.414958365849884, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0672\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'workclass' has levels not trained on: [Federal-gov, Local-gov, Never-worked]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [?, Adm-clerical, Armed-Forces, Farming-fishing, Handlers-cleaners, Machine-op-inspct, Priv-house-serv, Protective-serv, Tech-support, Transport-moving]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.6831530139103554\n",
      "Test accuracy: 0.8770007621951219\n",
      "\n",
      "Epoch 78\n",
      "Discriminator loss: -0.1922\n",
      "\n",
      "Epoch 79\n",
      "Discriminator loss: -0.5772\n",
      "epsilon = 8.755745563050132, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0527\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Farming-fishing, Handlers-cleaners, Machine-op-inspct, Other-service, Priv-house-serv, Protective-serv, Tech-support, Transport-moving]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7901854714064915\n",
      "Train accuracy: 0.6881761978361669\n",
      "Test accuracy: 0.8091653963414634\n",
      "\n",
      "Epoch 80\n",
      "Discriminator loss: -0.2039\n",
      "\n",
      "Epoch 81\n",
      "Discriminator loss: -0.3563\n",
      "epsilon = 9.096532760250382, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0599\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'workclass' has levels not trained on: [Federal-gov, Never-worked, Self-emp-inc]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [10th, 11th, 12th, 1st-4th, 5th-6th, 9th, Assoc-acdm, Doctorate, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Farming-fishing, Priv-house-serv, Protective-serv, Tech-support, Transport-moving]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [?, Cambodia, China, Columbia, Cuba, Dominican-Republic, Ecuador, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.6848918083462133\n",
      "Test accuracy: 0.749047256097561\n",
      "\n",
      "Epoch 82\n",
      "Discriminator loss: -0.1693\n",
      "\n",
      "Epoch 83\n",
      "Discriminator loss: -0.2612\n",
      "change\n",
      "epsilon = 9.367593805423091, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0424\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'race' has levels not trained on: [Amer-Indian-Eskimo, Asian-Pac-Islander, Other]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7841962905718701\n",
      "Train accuracy: 0.7161901081916537\n",
      "Test accuracy: 0.7595274390243902\n",
      "\n",
      "Epoch 84\n",
      "Discriminator loss: -0.357\n",
      "\n",
      "Epoch 85\n",
      "Discriminator loss: -0.0004\n",
      "epsilon = 9.62061768463569, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.055\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.741112828438949\n",
      "Test accuracy: 0.6948361280487805\n",
      "\n",
      "Epoch 86\n",
      "Discriminator loss: -0.2738\n",
      "\n",
      "Epoch 87\n",
      "Discriminator loss: -0.207\n",
      "change\n",
      "epsilon = 9.874292995960852, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0405\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "deeplearning prediction progress: |███████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [11th, 12th, 1st-4th, 5th-6th, 7th-8th, 9th, Assoc-acdm, Doctorate, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, Canada, China, Columbia, Cuba, Dominican-Republic, Ecuador, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7828438948995363\n",
      "Train accuracy: 0.6941653786707882\n",
      "Test accuracy: 0.794016768292683\n",
      "\n",
      "Epoch 88\n",
      "Discriminator loss: -0.2648\n",
      "\n",
      "Epoch 89\n",
      "Discriminator loss: -0.1376\n",
      "epsilon = 10.127968307286014, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0597\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [10th, 11th, 12th, 1st-4th, 5th-6th, 7th-8th, 9th, Assoc-acdm, Doctorate, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, China, Columbia, Cuba, Dominican-Republic, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7863214837712519\n",
      "Train accuracy: 0.6955177743431221\n",
      "Test accuracy: 0.7714367378048781\n",
      "\n",
      "Epoch 90\n",
      "Discriminator loss: -0.2408\n",
      "\n",
      "Epoch 91\n",
      "Discriminator loss: -0.1269\n",
      "epsilon = 10.381643618611177, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0443\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7291344667697064\n",
      "Test accuracy: 0.8364138719512195\n",
      "\n",
      "Epoch 92\n",
      "Discriminator loss: -0.386\n",
      "\n",
      "Epoch 93\n",
      "Discriminator loss: -0.102\n",
      "change\n",
      "epsilon = 10.63531892993634, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0455\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Farming-fishing, Machine-op-inspct, Priv-house-serv, Protective-serv, Tech-support, Transport-moving]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7832302936630603\n",
      "Train accuracy: 0.660548686244204\n",
      "Test accuracy: 0.672827743902439\n",
      "\n",
      "Epoch 94\n",
      "Discriminator loss: -0.2382\n",
      "\n",
      "Epoch 95\n",
      "Discriminator loss: -0.2939\n",
      "change\n",
      "epsilon = 10.888994241261502, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0718\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'workclass' has levels not trained on: [Never-worked, Self-emp-inc]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, China, Columbia, Cuba, Dominican-Republic, Ecuador, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.6389103554868625\n",
      "Test accuracy: 0.604420731707317\n",
      "\n",
      "Epoch 96\n",
      "Discriminator loss: -0.1724\n",
      "\n",
      "Epoch 97\n",
      "Discriminator loss: -0.1939\n",
      "change\n",
      "epsilon = 11.142669552586664, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0483\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'education' has levels not trained on: [10th, 11th, 12th, 1st-4th, 5th-6th, 9th, Assoc-acdm, Assoc-voc, Doctorate, Preschool, Prof-school]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'relationship' has levels not trained on: [Other-relative, Own-child]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7836166924265843\n",
      "Train accuracy: 0.5850077279752705\n",
      "Test accuracy: 0.6919778963414634\n",
      "\n",
      "Epoch 98\n",
      "Discriminator loss: -0.1551\n",
      "\n",
      "Epoch 99\n",
      "Discriminator loss: -0.1275\n",
      "epsilon = 11.396344863911827, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0374\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7822642967542504\n",
      "Train accuracy: 0.6918469860896446\n",
      "Test accuracy: 0.659108231707317\n",
      "\n",
      "Epoch 100\n",
      "Discriminator loss: -0.1208\n",
      "\n",
      "Epoch 101\n",
      "Discriminator loss: -0.0261\n",
      "epsilon = 11.650020175236989, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0327\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'occupation' has levels not trained on: [Armed-Forces, Farming-fishing, Other-service, Priv-house-serv, Protective-serv, Tech-support, Transport-moving]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'relationship' has levels not trained on: [Other-relative, Unmarried]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785355486862442\n",
      "Train accuracy: 0.7044049459041731\n",
      "Test accuracy: 0.8004954268292683\n",
      "\n",
      "Epoch 102\n",
      "Discriminator loss: -0.1054\n",
      "\n",
      "Epoch 103\n",
      "Discriminator loss: -0.3656\n",
      "epsilon = 11.903695486562151, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0653\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'workclass' has levels not trained on: [Federal-gov, Never-worked, State-gov]\n",
      "  warnings.warn(w)\n",
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, China, Columbia, Cuba, Dominican-Republic, Ecuador, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785355486862442\n",
      "Train accuracy: 0.6829598145285936\n",
      "Test accuracy: 0.6930259146341463\n",
      "\n",
      "Epoch 104\n",
      "Discriminator loss: -0.1594\n",
      "\n",
      "Epoch 105\n",
      "Discriminator loss: -0.2067\n",
      "change\n",
      "epsilon = 12.157370797887314, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0449\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7134853168469861\n",
      "Test accuracy: 0.7661013719512195\n",
      "\n",
      "Epoch 106\n",
      "Discriminator loss: -0.0599\n",
      "\n",
      "Epoch 107\n",
      "Discriminator loss: -0.2263\n",
      "epsilon = 12.411046109212476, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0323\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |█████████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7832302936630603\n",
      "Train accuracy: 0.7202472952086554\n",
      "Test accuracy: 0.7562881097560976\n",
      "\n",
      "Epoch 108\n",
      "Discriminator loss: -0.171\n",
      "\n",
      "Epoch 109\n",
      "Discriminator loss: -0.2199\n",
      "epsilon = 12.66472142053765, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0354\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7841962905718701\n",
      "Train accuracy: 0.6373647604327666\n",
      "Test accuracy: 0.6019435975609756\n",
      "\n",
      "Epoch 110\n",
      "Discriminator loss: -0.1293\n",
      "\n",
      "Epoch 111\n",
      "Discriminator loss: -0.0353\n",
      "epsilon = 12.918396731862885, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0418\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7836166924265843\n",
      "Train accuracy: 0.7244976816074189\n",
      "Test accuracy: 0.8202172256097561\n",
      "\n",
      "Epoch 112\n",
      "Discriminator loss: -0.198\n",
      "\n",
      "Epoch 113\n",
      "Discriminator loss: -0.0629\n",
      "epsilon = 13.17207204318812, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.034\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7834234930448223\n",
      "Train accuracy: 0.7443972179289027\n",
      "Test accuracy: 0.8259336890243902\n",
      "\n",
      "Epoch 114\n",
      "Discriminator loss: -0.1471\n",
      "\n",
      "Epoch 115\n",
      "Discriminator loss: -0.157\n",
      "change\n",
      "epsilon = 13.425747354513355, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0473\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7845826893353941\n",
      "Train accuracy: 0.7167697063369397\n",
      "Test accuracy: 0.8384146341463414\n",
      "\n",
      "Epoch 116\n",
      "Discriminator loss: -0.0743\n",
      "\n",
      "Epoch 117\n",
      "Discriminator loss: -0.1284\n",
      "change\n",
      "epsilon = 13.67942266583859, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0446\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7474884080370943\n",
      "Test accuracy: 0.811451981707317\n",
      "\n",
      "Epoch 118\n",
      "Discriminator loss: -0.1783\n",
      "\n",
      "Epoch 119\n",
      "Discriminator loss: -0.1743\n",
      "change\n",
      "epsilon = 13.933097977163825, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0326\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7828438948995363\n",
      "Train accuracy: 0.7610123647604328\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test accuracy: 0.7558117378048781\n",
      "\n",
      "Epoch 120\n",
      "Discriminator loss: -0.1254\n",
      "\n",
      "Epoch 121\n",
      "Discriminator loss: -0.1018\n",
      "epsilon = 14.18677328848906, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0367\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785548686244204\n",
      "Train accuracy: 0.7461360123647605\n",
      "Test accuracy: 0.770579268292683\n",
      "\n",
      "Epoch 122\n",
      "Discriminator loss: -0.103\n",
      "\n",
      "Epoch 123\n",
      "Discriminator loss: -0.171\n",
      "epsilon = 14.440448599814296, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0342\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7834234930448223\n",
      "Train accuracy: 0.741306027820711\n",
      "Test accuracy: 0.8360327743902439\n",
      "\n",
      "Epoch 124\n",
      "Discriminator loss: -0.1441\n",
      "\n",
      "Epoch 125\n",
      "Discriminator loss: -0.07\n",
      "change\n",
      "epsilon = 14.694123911139531, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0299\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7843894899536321\n",
      "Train accuracy: 0.7030525502318392\n",
      "Test accuracy: 0.7183689024390244\n",
      "\n",
      "Epoch 126\n",
      "Discriminator loss: -0.0883\n",
      "\n",
      "Epoch 127\n",
      "Discriminator loss: -0.1662\n",
      "epsilon = 14.947799222464766, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0345\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7824574961360123\n",
      "Train accuracy: 0.7345440494590417\n",
      "Test accuracy: 0.8082126524390244\n",
      "\n",
      "Epoch 128\n",
      "Discriminator loss: -0.1275\n",
      "\n",
      "Epoch 129\n",
      "Discriminator loss: -0.1568\n",
      "epsilon = 15.201474533790002, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0385\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7849690880989181\n",
      "Train accuracy: 0.7076893353941267\n",
      "Test accuracy: 0.7502858231707317\n",
      "\n",
      "Epoch 130\n",
      "Discriminator loss: -0.1489\n",
      "\n",
      "Epoch 131\n",
      "Discriminator loss: -0.1412\n",
      "epsilon = 15.455149845115237, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0374\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7828438948995363\n",
      "Train accuracy: 0.7436244204018547\n",
      "Test accuracy: 0.7369474085365854\n",
      "\n",
      "Epoch 132\n",
      "Discriminator loss: -0.0415\n",
      "\n",
      "Epoch 133\n",
      "Discriminator loss: -0.1732\n",
      "change\n",
      "epsilon = 15.708825156440472, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.029\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7841962905718701\n",
      "Train accuracy: 0.7279752704791345\n",
      "Test accuracy: 0.8127858231707317\n",
      "\n",
      "Epoch 134\n",
      "Discriminator loss: -0.0891\n",
      "\n",
      "Epoch 135\n",
      "Discriminator loss: -0.2965\n",
      "change\n",
      "epsilon = 15.962500467765707, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0463\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7830370942812983\n",
      "Train accuracy: 0.7343508500772797\n",
      "Test accuracy: 0.643578506097561\n",
      "\n",
      "Epoch 136\n",
      "Discriminator loss: -0.1333\n",
      "\n",
      "Epoch 137\n",
      "Discriminator loss: -0.132\n",
      "epsilon = 16.216175777683002, delta = 9.518370454978108e-05\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wasserstein distance: 0.0433\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7824574961360123\n",
      "Train accuracy: 0.6991885625965997\n",
      "Test accuracy: 0.6267149390243902\n",
      "\n",
      "Epoch 138\n",
      "Discriminator loss: -0.1407\n",
      "\n",
      "Epoch 139\n",
      "Discriminator loss: -0.0637\n",
      "epsilon = 16.469199655487735, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0443\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785741885625966\n",
      "Train accuracy: 0.7304868624420402\n",
      "Test accuracy: 0.7734375\n",
      "\n",
      "Epoch 140\n",
      "Discriminator loss: -0.1262\n",
      "\n",
      "Epoch 141\n",
      "Discriminator loss: -0.0769\n",
      "change\n",
      "epsilon = 16.72287496681297, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0349\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7840030911901081\n",
      "Train accuracy: 0.740533230293663\n",
      "Test accuracy: 0.8306974085365854\n",
      "\n",
      "Epoch 142\n",
      "Discriminator loss: -0.0608\n",
      "\n",
      "Epoch 143\n",
      "Discriminator loss: -0.0653\n",
      "epsilon = 16.976550278138205, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0301\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [?, Cambodia, Canada, China, Columbia, Cuba, Dominican-Republic, Ecuador, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7849690880989181\n",
      "Train accuracy: 0.7455564142194745\n",
      "Test accuracy: 0.8035442073170732\n",
      "\n",
      "Epoch 144\n",
      "Discriminator loss: -0.1194\n",
      "\n",
      "Epoch 145\n",
      "Discriminator loss: -0.1071\n",
      "change\n",
      "epsilon = 17.23022558946344, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0339\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785355486862442\n",
      "Train accuracy: 0.740919629057187\n",
      "Test accuracy: 0.715891768292683\n",
      "\n",
      "Epoch 146\n",
      "Discriminator loss: -0.1754\n",
      "\n",
      "Epoch 147\n",
      "Discriminator loss: -0.1354\n",
      "change\n",
      "epsilon = 17.483900900788676, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0281\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, Canada, China, Columbia, Cuba, Dominican-Republic, Ecuador, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Mexico, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7828438948995363\n",
      "Train accuracy: 0.7447836166924265\n",
      "Test accuracy: 0.8341272865853658\n",
      "\n",
      "Epoch 148\n",
      "Discriminator loss: -0.1551\n",
      "\n",
      "Epoch 149\n",
      "Discriminator loss: -0.1583\n",
      "epsilon = 17.73757621211391, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.032\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7453632148377125\n",
      "Test accuracy: 0.7434260670731707\n",
      "\n",
      "Epoch 150\n",
      "Discriminator loss: -0.1375\n",
      "\n",
      "Epoch 151\n",
      "Discriminator loss: -0.1197\n",
      "epsilon = 17.99125152203121, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0262\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7818778979907264\n",
      "Train accuracy: 0.7376352395672334\n",
      "Test accuracy: 0.700171493902439\n",
      "\n",
      "Epoch 152\n",
      "Discriminator loss: -0.0901\n",
      "\n",
      "Epoch 153\n",
      "Discriminator loss: -0.1628\n",
      "epsilon = 18.244926833356445, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0446\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7841962905718701\n",
      "Train accuracy: 0.741499227202473\n",
      "Test accuracy: 0.7738185975609756\n",
      "\n",
      "Epoch 154\n",
      "Discriminator loss: -0.0861\n",
      "\n",
      "Epoch 155\n",
      "Discriminator loss: -0.1347\n",
      "change\n",
      "epsilon = 18.443478027632075, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0303\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785548686244204\n",
      "Train accuracy: 0.7376352395672334\n",
      "Test accuracy: 0.8052591463414634\n",
      "\n",
      "Epoch 156\n",
      "Discriminator loss: -0.1776\n",
      "\n",
      "Epoch 157\n",
      "Discriminator loss: -0.1097\n",
      "epsilon = 18.611345734772662, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0228\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7822642967542504\n",
      "Train accuracy: 0.7304868624420402\n",
      "Test accuracy: 0.8256478658536586\n",
      "\n",
      "Epoch 158\n",
      "Discriminator loss: -0.1185\n",
      "\n",
      "Epoch 159\n",
      "Discriminator loss: -0.0846\n",
      "epsilon = 18.779213441913257, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0415\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7832302936630603\n",
      "Train accuracy: 0.7445904173106646\n",
      "Test accuracy: 0.7904916158536586\n",
      "\n",
      "Epoch 160\n",
      "Discriminator loss: -0.131\n",
      "\n",
      "Epoch 161\n",
      "Discriminator loss: -0.1519\n",
      "change\n",
      "epsilon = 18.947081149053844, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0297\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7107805255023184\n",
      "Test accuracy: 0.6185213414634146\n",
      "\n",
      "Epoch 162\n",
      "Discriminator loss: -0.1015\n",
      "\n",
      "Epoch 163\n",
      "Discriminator loss: -0.1518\n",
      "change\n",
      "epsilon = 19.114948856194438, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0276\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7840030911901081\n",
      "Train accuracy: 0.7387944358578052\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test accuracy: 0.6917873475609756\n",
      "\n",
      "Epoch 164\n",
      "Discriminator loss: -0.1596\n",
      "\n",
      "Epoch 165\n",
      "Discriminator loss: -0.1334\n",
      "epsilon = 19.282816563335025, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0274\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785548686244204\n",
      "Train accuracy: 0.7387944358578052\n",
      "Test accuracy: 0.796875\n",
      "\n",
      "Epoch 166\n",
      "Discriminator loss: -0.0575\n",
      "\n",
      "Epoch 167\n",
      "Discriminator loss: -0.0647\n",
      "change\n",
      "epsilon = 19.45068427047562, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0265\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7521251931993818\n",
      "Test accuracy: 0.6702553353658537\n",
      "\n",
      "Epoch 168\n",
      "Discriminator loss: -0.0979\n",
      "\n",
      "Epoch 169\n",
      "Discriminator loss: -0.0496\n",
      "epsilon = 19.618551977616207, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.024\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, China, Columbia, Cuba, Dominican-Republic, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7841962905718701\n",
      "Train accuracy: 0.759273570324575\n",
      "Test accuracy: 0.7212271341463414\n",
      "\n",
      "Epoch 170\n",
      "Discriminator loss: -0.1231\n",
      "\n",
      "Epoch 171\n",
      "Discriminator loss: -0.1216\n",
      "epsilon = 19.7864196847568, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0245\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, Canada, China, Columbia, Cuba, Dominican-Republic, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Thailand, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7847758887171561\n",
      "Train accuracy: 0.7581143740340031\n",
      "Test accuracy: 0.7424733231707317\n",
      "\n",
      "Epoch 172\n",
      "Discriminator loss: -0.1106\n",
      "\n",
      "Epoch 173\n",
      "Discriminator loss: -0.0782\n",
      "change\n",
      "epsilon = 19.954287391897388, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0232\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7836166924265843\n",
      "Train accuracy: 0.7561823802163833\n",
      "Test accuracy: 0.7555259146341463\n",
      "\n",
      "Epoch 174\n",
      "Discriminator loss: -0.0592\n",
      "\n",
      "Epoch 175\n",
      "Discriminator loss: -0.0975\n",
      "epsilon = 20.122155099037982, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0254\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7826506955177743\n",
      "Train accuracy: 0.7355100463678517\n",
      "Test accuracy: 0.803639481707317\n",
      "\n",
      "Epoch 176\n",
      "Discriminator loss: -0.1591\n",
      "\n",
      "Epoch 177\n",
      "Discriminator loss: -0.1151\n",
      "change\n",
      "epsilon = 20.29002280617857, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0263\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7841962905718701\n",
      "Train accuracy: 0.7428516228748068\n",
      "Test accuracy: 0.7625762195121951\n",
      "\n",
      "Epoch 178\n",
      "Discriminator loss: -0.1392\n",
      "\n",
      "Epoch 179\n",
      "Discriminator loss: -0.1008\n",
      "epsilon = 20.457890513319164, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0207\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7832302936630603\n",
      "Train accuracy: 0.749420401854714\n",
      "Test accuracy: 0.7474275914634146\n",
      "\n",
      "Epoch 180\n",
      "Discriminator loss: -0.0848\n",
      "\n",
      "Epoch 181\n",
      "Discriminator loss: -0.0998\n",
      "change\n",
      "epsilon = 20.62575822045975, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0227\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7834234930448223\n",
      "Train accuracy: 0.7538639876352395\n",
      "Test accuracy: 0.8139291158536586\n",
      "\n",
      "Epoch 182\n",
      "Discriminator loss: -0.0708\n",
      "\n",
      "Epoch 183\n",
      "Discriminator loss: -0.2211\n",
      "epsilon = 20.793625927600345, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0343\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7828438948995363\n",
      "Train accuracy: 0.7198608964451314\n",
      "Test accuracy: 0.6952172256097561\n",
      "\n",
      "Epoch 184\n",
      "Discriminator loss: -0.1025\n",
      "\n",
      "Epoch 185\n",
      "Discriminator loss: -0.098\n",
      "change\n",
      "epsilon = 20.961493634740933, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.021\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7841962905718701\n",
      "Train accuracy: 0.7511591962905718\n",
      "Test accuracy: 0.8044969512195121\n",
      "\n",
      "Epoch 186\n",
      "Discriminator loss: -0.1167\n",
      "\n",
      "Epoch 187\n",
      "Discriminator loss: -0.1105\n",
      "epsilon = 21.129361341881527, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0264\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7550231839258115\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test accuracy: 0.782202743902439\n",
      "\n",
      "Epoch 188\n",
      "Discriminator loss: -0.1054\n",
      "\n",
      "Epoch 189\n",
      "Discriminator loss: -0.1666\n",
      "epsilon = 21.297229049022114, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0324\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |██| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7469088098918083\n",
      "Test accuracy: 0.7227515243902439\n",
      "\n",
      "Epoch 190\n",
      "Discriminator loss: -0.0701\n",
      "\n",
      "Epoch 191\n",
      "Discriminator loss: -0.1148\n",
      "change\n",
      "epsilon = 21.46509675616271, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0177\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.78612828438949\n",
      "Train accuracy: 0.759080370942813\n",
      "Test accuracy: 0.7485708841463414\n",
      "\n",
      "Epoch 192\n",
      "Discriminator loss: -0.1014\n",
      "\n",
      "Epoch 193\n",
      "Discriminator loss: -0.0922\n",
      "change\n",
      "epsilon = 21.632964463303296, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0269\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7843894899536321\n",
      "Train accuracy: 0.7606259659969088\n",
      "Test accuracy: 0.766577743902439\n",
      "\n",
      "Epoch 194\n",
      "Discriminator loss: -0.0625\n",
      "\n",
      "Epoch 195\n",
      "Discriminator loss: -0.1192\n",
      "change\n",
      "epsilon = 21.80083217044389, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0215\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7841962905718701\n",
      "Train accuracy: 0.7542503863987635\n",
      "Test accuracy: 0.8048780487804879\n",
      "\n",
      "Epoch 196\n",
      "Discriminator loss: -0.1037\n",
      "\n",
      "Epoch 197\n",
      "Discriminator loss: -0.1191\n",
      "epsilon = 21.968699877584477, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0237\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, China, Columbia, Cuba, Dominican-Republic, Ecuador, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Mexico, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7870942812982998\n",
      "Train accuracy: 0.7565687789799073\n",
      "Test accuracy: 0.7639100609756098\n",
      "\n",
      "Epoch 198\n",
      "Discriminator loss: -0.1502\n",
      "\n",
      "Epoch 199\n",
      "Discriminator loss: -0.1209\n",
      "change\n",
      "epsilon = 22.13656758472507, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0218\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7840030911901081\n",
      "Train accuracy: 0.7602395672333848\n",
      "Test accuracy: 0.7761051829268293\n",
      "\n",
      "Epoch 200\n",
      "Discriminator loss: -0.1421\n",
      "\n",
      "Epoch 201\n",
      "Discriminator loss: -0.0717\n",
      "epsilon = 22.30443529186566, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0208\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7557959814528593\n",
      "Test accuracy: 0.7376143292682927\n",
      "\n",
      "Epoch 202\n",
      "Discriminator loss: -0.1059\n",
      "\n",
      "Epoch 203\n",
      "Discriminator loss: -0.1719\n",
      "change\n",
      "epsilon = 22.472302999006253, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0256\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7828438948995363\n",
      "Train accuracy: 0.7571483771251932\n",
      "Test accuracy: 0.7029344512195121\n",
      "\n",
      "Epoch 204\n",
      "Discriminator loss: -0.0901\n",
      "\n",
      "Epoch 205\n",
      "Discriminator loss: -0.1595\n",
      "change\n",
      "epsilon = 22.64017070614684, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0258\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7787867078825348\n",
      "Train accuracy: 0.7357032457496137\n",
      "Test accuracy: 0.7613376524390244\n",
      "\n",
      "Epoch 206\n",
      "Discriminator loss: -0.1128\n",
      "\n",
      "Epoch 207\n",
      "Discriminator loss: -0.1127\n",
      "epsilon = 22.808038413287434, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0211\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785355486862442\n",
      "Train accuracy: 0.759080370942813\n",
      "Test accuracy: 0.7538109756097561\n",
      "\n",
      "Epoch 208\n",
      "Discriminator loss: -0.0882\n",
      "\n",
      "Epoch 209\n",
      "Discriminator loss: -0.1298\n",
      "epsilon = 22.97590612042802, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0212\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7816846986089645\n",
      "Train accuracy: 0.7664219474497682\n",
      "Test accuracy: 0.7521913109756098\n",
      "\n",
      "Epoch 210\n",
      "Discriminator loss: -0.0669\n",
      "\n",
      "Epoch 211\n",
      "Discriminator loss: -0.1017\n",
      "change\n",
      "epsilon = 23.143773827568616, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0203\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7830370942812983\n",
      "Train accuracy: 0.7612055641421948\n",
      "Test accuracy: 0.7560975609756098\n",
      "\n",
      "Epoch 212\n",
      "Discriminator loss: -0.0939\n",
      "\n",
      "Epoch 213\n",
      "Discriminator loss: -0.0837\n",
      "change\n",
      "epsilon = 23.311641534709203, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0174\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7849690880989181\n",
      "Train accuracy: 0.7619783616692427\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test accuracy: 0.7621951219512195\n",
      "\n",
      "Epoch 214\n",
      "Discriminator loss: -0.0585\n",
      "\n",
      "Epoch 215\n",
      "Discriminator loss: -0.0837\n",
      "epsilon = 23.479509241849797, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0238\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7849690880989181\n",
      "Train accuracy: 0.7532843894899537\n",
      "Test accuracy: 0.7946836890243902\n",
      "\n",
      "Epoch 216\n",
      "Discriminator loss: -0.1205\n",
      "\n",
      "Epoch 217\n",
      "Discriminator loss: -0.0634\n",
      "change\n",
      "epsilon = 23.647376948990384, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0247\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7836166924265843\n",
      "Train accuracy: 0.7523183925811437\n",
      "Test accuracy: 0.7989710365853658\n",
      "\n",
      "Epoch 218\n",
      "Discriminator loss: -0.1163\n",
      "\n",
      "Epoch 219\n",
      "Discriminator loss: -0.1404\n",
      "epsilon = 23.81524465613098, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0208\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7438176197836167\n",
      "Test accuracy: 0.7848704268292683\n",
      "\n",
      "Epoch 220\n",
      "Discriminator loss: -0.1258\n",
      "\n",
      "Epoch 221\n",
      "Discriminator loss: -0.0898\n",
      "change\n",
      "epsilon = 23.983112363271566, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.027\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7820710973724884\n",
      "Train accuracy: 0.7069165378670789\n",
      "Test accuracy: 0.8084032012195121\n",
      "\n",
      "Epoch 222\n",
      "Discriminator loss: -0.116\n",
      "\n",
      "Epoch 223\n",
      "Discriminator loss: -0.091\n",
      "epsilon = 24.15098007041216, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0274\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7807187017001546\n",
      "Train accuracy: 0.7552163833075735\n",
      "Test accuracy: 0.6918826219512195\n",
      "\n",
      "Epoch 224\n",
      "Discriminator loss: -0.0754\n",
      "\n",
      "Epoch 225\n",
      "Discriminator loss: -0.0841\n",
      "epsilon = 24.318847777552747, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0249\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7656491499227203\n",
      "Test accuracy: 0.7857278963414634\n",
      "\n",
      "Epoch 226\n",
      "Discriminator loss: -0.1068\n",
      "\n",
      "Epoch 227\n",
      "Discriminator loss: -0.0845\n",
      "change\n",
      "epsilon = 24.48671548469334, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.018\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\i061767\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h2o\\job.py:69: UserWarning: Test/Validation dataset column 'native-country' has levels not trained on: [Cambodia, Canada, China, Columbia, Cuba, Dominican-Republic, El-Salvador, England, France, Germany, Greece, Guatemala, Haiti, Honduras, Hong, Hungary, India, Iran, Ireland, Italy, Jamaica, Japan, Laos, Nicaragua, Outlying-US(Guam-USVI-etc), Peru, Philippines, Poland, Portugal, Puerto-Rico, South, Taiwan, Vietnam, Yugoslavia]\n",
      "  warnings.warn(w)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7834234930448223\n",
      "Train accuracy: 0.7631375579598145\n",
      "Test accuracy: 0.6844512195121951\n",
      "\n",
      "Epoch 228\n",
      "Discriminator loss: -0.0878\n",
      "\n",
      "Epoch 229\n",
      "Discriminator loss: -0.1154\n",
      "change\n",
      "epsilon = 24.65458319183393, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0243\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7608191653786708\n",
      "Test accuracy: 0.778296493902439\n",
      "\n",
      "Epoch 230\n",
      "Discriminator loss: -0.1228\n",
      "\n",
      "Epoch 231\n",
      "Discriminator loss: -0.0941\n",
      "epsilon = 24.822450898974523, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0228\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7838098918083463\n",
      "Train accuracy: 0.759466769706337\n",
      "Test accuracy: 0.7569550304878049\n",
      "\n",
      "Epoch 232\n",
      "Discriminator loss: -0.0763\n",
      "\n",
      "Epoch 233\n",
      "Discriminator loss: -0.0915\n",
      "change\n",
      "epsilon = 24.99031860611511, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0245\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7378284389489954\n",
      "Test accuracy: 0.7698170731707317\n",
      "\n",
      "Epoch 234\n",
      "Discriminator loss: -0.1073\n",
      "\n",
      "Epoch 235\n",
      "Discriminator loss: -0.1144\n",
      "epsilon = 25.158186313255705, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0189\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7841962905718701\n",
      "Train accuracy: 0.7424652241112828\n",
      "Test accuracy: 0.7575266768292683\n",
      "\n",
      "Epoch 236\n",
      "Discriminator loss: -0.1037\n",
      "\n",
      "Epoch 237\n",
      "Discriminator loss: -0.0875\n",
      "change\n",
      "epsilon = 25.326054020396697, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0179\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7832302936630603\n",
      "Train accuracy: 0.749806800618238\n",
      "Test accuracy: 0.7408536585365854\n",
      "\n",
      "Epoch 238\n",
      "Discriminator loss: -0.0765\n",
      "\n",
      "Epoch 239\n",
      "Discriminator loss: -0.1041\n",
      "change\n",
      "epsilon = 25.493921727538307, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0197\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7836166924265843\n",
      "Train accuracy: 0.7272024729520865\n",
      "Test accuracy: 0.7762004573170732\n",
      "\n",
      "Epoch 240\n",
      "Discriminator loss: -0.0753\n",
      "\n",
      "Epoch 241\n",
      "Discriminator loss: -0.126\n",
      "epsilon = 25.661789434679918, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.015\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7865146831530139\n",
      "Train accuracy: 0.7461360123647605\n",
      "Test accuracy: 0.7635289634146342\n",
      "\n",
      "Epoch 242\n",
      "Discriminator loss: -0.0997\n",
      "\n",
      "Epoch 243\n",
      "Discriminator loss: -0.075\n",
      "change\n",
      "epsilon = 25.829657141821528, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0225\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7836166924265843\n",
      "Train accuracy: 0.7418856259659969\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test accuracy: 0.749047256097561\n",
      "\n",
      "Epoch 244\n",
      "Discriminator loss: -0.1147\n",
      "\n",
      "Epoch 245\n",
      "Discriminator loss: -0.0822\n",
      "epsilon = 25.99752484896314, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0213\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7208268933539412\n",
      "Test accuracy: 0.8101181402439024\n",
      "\n",
      "Epoch 246\n",
      "Discriminator loss: -0.097\n",
      "\n",
      "Epoch 247\n",
      "Discriminator loss: -0.0917\n",
      "change\n",
      "epsilon = 26.16539255610475, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0183\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7838098918083463\n",
      "Train accuracy: 0.7289412673879444\n",
      "Test accuracy: 0.7419016768292683\n",
      "\n",
      "Epoch 248\n",
      "Discriminator loss: -0.0988\n",
      "\n",
      "Epoch 249\n",
      "Discriminator loss: -0.063\n",
      "epsilon = 26.33326026324636, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.02\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.740146831530139\n",
      "Test accuracy: 0.7291349085365854\n",
      "\n",
      "Epoch 250\n",
      "Discriminator loss: -0.1076\n",
      "\n",
      "Epoch 251\n",
      "Discriminator loss: -0.0721\n",
      "change\n",
      "epsilon = 26.50112797038797, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0165\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7809119010819165\n",
      "Train accuracy: 0.7372488408037094\n",
      "Test accuracy: 0.7688643292682927\n",
      "\n",
      "Epoch 252\n",
      "Discriminator loss: -0.0614\n",
      "\n",
      "Epoch 253\n",
      "Discriminator loss: -0.0806\n",
      "epsilon = 26.66899567752958, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0203\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.73241885625966\n",
      "Test accuracy: 0.702172256097561\n",
      "\n",
      "Epoch 254\n",
      "Discriminator loss: -0.0583\n",
      "\n",
      "Epoch 255\n",
      "Discriminator loss: -0.0948\n",
      "change\n",
      "epsilon = 26.83686338467119, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0173\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7384080370942813\n",
      "Test accuracy: 0.7644817073170732\n",
      "\n",
      "Epoch 256\n",
      "Discriminator loss: -0.1053\n",
      "\n",
      "Epoch 257\n",
      "Discriminator loss: -0.0957\n",
      "change\n",
      "epsilon = 27.0047310918128, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.019\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7845826893353941\n",
      "Train accuracy: 0.7554095826893354\n",
      "Test accuracy: 0.784203506097561\n",
      "\n",
      "Epoch 258\n",
      "Discriminator loss: -0.0946\n",
      "\n",
      "Epoch 259\n",
      "Discriminator loss: -0.0611\n",
      "change\n",
      "epsilon = 27.17259879895441, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0231\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7832302936630603\n",
      "Train accuracy: 0.749613601236476\n",
      "Test accuracy: 0.7999237804878049\n",
      "\n",
      "Epoch 260\n",
      "Discriminator loss: -0.0529\n",
      "\n",
      "Epoch 261\n",
      "Discriminator loss: -0.0936\n",
      "change\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epsilon = 27.34046650609602, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0203\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7828438948995363\n",
      "Train accuracy: 0.7540571870170015\n",
      "Test accuracy: 0.7877286585365854\n",
      "\n",
      "Epoch 262\n",
      "Discriminator loss: -0.1136\n",
      "\n",
      "Epoch 263\n",
      "Discriminator loss: -0.0915\n",
      "epsilon = 27.50833421323763, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0187\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.73241885625966\n",
      "Test accuracy: 0.7096989329268293\n",
      "\n",
      "Epoch 264\n",
      "Discriminator loss: -0.1334\n",
      "\n",
      "Epoch 265\n",
      "Discriminator loss: -0.0968\n",
      "epsilon = 27.676201920379242, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0169\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7963678516228748\n",
      "Train accuracy: 0.7565687789799073\n",
      "Test accuracy: 0.7811547256097561\n",
      "\n",
      "Epoch 266\n",
      "Discriminator loss: -0.1205\n",
      "\n",
      "Epoch 267\n",
      "Discriminator loss: -0.083\n",
      "epsilon = 27.844069627520852, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.019\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7849690880989181\n",
      "Train accuracy: 0.7461360123647605\n",
      "Test accuracy: 0.8106897865853658\n",
      "\n",
      "Epoch 268\n",
      "Discriminator loss: -0.0913\n",
      "\n",
      "Epoch 269\n",
      "Discriminator loss: -0.108\n",
      "change\n",
      "epsilon = 28.011937334662463, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.016\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7621715610510046\n",
      "Test accuracy: 0.7369474085365854\n",
      "\n",
      "Epoch 270\n",
      "Discriminator loss: -0.1091\n",
      "\n",
      "Epoch 271\n",
      "Discriminator loss: -0.1031\n",
      "change\n",
      "epsilon = 28.179805041804073, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0183\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7608191653786708\n",
      "Test accuracy: 0.7721036585365854\n",
      "\n",
      "Epoch 272\n",
      "Discriminator loss: -0.0806\n",
      "\n",
      "Epoch 273\n",
      "Discriminator loss: -0.0876\n",
      "epsilon = 28.347672748945683, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0189\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7836166924265843\n",
      "Train accuracy: 0.7619783616692427\n",
      "Test accuracy: 0.762671493902439\n",
      "\n",
      "Epoch 274\n",
      "Discriminator loss: -0.1287\n",
      "\n",
      "Epoch 275\n",
      "Discriminator loss: -0.1084\n",
      "change\n",
      "epsilon = 28.515540456087294, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0188\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7838098918083463\n",
      "Train accuracy: 0.7525115919629057\n",
      "Test accuracy: 0.7884908536585366\n",
      "\n",
      "Epoch 276\n",
      "Discriminator loss: -0.1577\n",
      "\n",
      "Epoch 277\n",
      "Discriminator loss: -0.1259\n",
      "epsilon = 28.683408163228904, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0166\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7832302936630603\n",
      "Train accuracy: 0.7461360123647605\n",
      "Test accuracy: 0.8013528963414634\n",
      "\n",
      "Epoch 278\n",
      "Discriminator loss: -0.0723\n",
      "\n",
      "Epoch 279\n",
      "Discriminator loss: -0.0828\n",
      "change\n",
      "epsilon = 28.851275870370515, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.017\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7845826893353941\n",
      "Train accuracy: 0.7552163833075735\n",
      "Test accuracy: 0.7671493902439024\n",
      "\n",
      "Epoch 280\n",
      "Discriminator loss: -0.0779\n",
      "\n",
      "Epoch 281\n",
      "Discriminator loss: -0.0686\n",
      "change\n",
      "epsilon = 29.019143577512125, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0172\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7832302936630603\n",
      "Train accuracy: 0.7557959814528593\n",
      "Test accuracy: 0.7011242378048781\n",
      "\n",
      "Epoch 282\n",
      "Discriminator loss: -0.1203\n",
      "\n",
      "Epoch 283\n",
      "Discriminator loss: -0.0472\n",
      "epsilon = 29.187011284653735, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0178\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7832302936630603\n",
      "Train accuracy: 0.7465224111282844\n",
      "Test accuracy: 0.7296112804878049\n",
      "\n",
      "Epoch 284\n",
      "Discriminator loss: -0.1009\n",
      "\n",
      "Epoch 285\n",
      "Discriminator loss: -0.1068\n",
      "epsilon = 29.354878991795346, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0168\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7832302936630603\n",
      "Train accuracy: 0.7563755795981453\n",
      "Test accuracy: 0.7408536585365854\n",
      "\n",
      "Epoch 286\n",
      "Discriminator loss: -0.0899\n",
      "\n",
      "Epoch 287\n",
      "Discriminator loss: -0.085\n",
      "epsilon = 29.522746698936956, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0173\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785741885625966\n",
      "Train accuracy: 0.7459428129829985\n",
      "Test accuracy: 0.7844893292682927\n",
      "\n",
      "Epoch 288\n",
      "Discriminator loss: -0.1159\n",
      "\n",
      "Epoch 289\n",
      "Discriminator loss: -0.0823\n",
      "change\n",
      "epsilon = 29.690614406078566, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0156\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7791731066460588\n",
      "Train accuracy: 0.7525115919629057\n",
      "Test accuracy: 0.7179878048780488\n",
      "\n",
      "Epoch 290\n",
      "Discriminator loss: -0.0641\n",
      "\n",
      "Epoch 291\n",
      "Discriminator loss: -0.0891\n",
      "epsilon = 29.858482113220177, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0187\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7847758887171561\n",
      "Train accuracy: 0.741692426584235\n",
      "Test accuracy: 0.7345655487804879\n",
      "\n",
      "Epoch 292\n",
      "Discriminator loss: -0.0539\n",
      "\n",
      "Epoch 293\n",
      "Discriminator loss: -0.0964\n",
      "epsilon = 30.026349820361787, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0179\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7838098918083463\n",
      "Train accuracy: 0.7389876352395672\n",
      "Test accuracy: 0.7375190548780488\n",
      "\n",
      "Epoch 294\n",
      "Discriminator loss: -0.0982\n",
      "\n",
      "Epoch 295\n",
      "Discriminator loss: -0.0914\n",
      "change\n",
      "epsilon = 30.194217527503397, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0157\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "glm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7836166924265843\n",
      "Train accuracy: 0.7345440494590417\n",
      "Test accuracy: 0.7406631097560976\n",
      "\n",
      "Epoch 296\n",
      "Discriminator loss: -0.1344\n",
      "\n",
      "Epoch 297\n",
      "Discriminator loss: -0.1629\n",
      "change\n",
      "epsilon = 30.362085234645008, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0223\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785548686244204\n",
      "Train accuracy: 0.7389876352395672\n",
      "Test accuracy: 0.739233993902439\n",
      "\n",
      "Epoch 298\n",
      "Discriminator loss: -0.0903\n",
      "\n",
      "Epoch 299\n",
      "Discriminator loss: -0.1049\n",
      "change\n",
      "epsilon = 30.52995294178662, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0174\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7838098918083463\n",
      "Train accuracy: 0.7556027820710973\n",
      "Test accuracy: 0.6861661585365854\n",
      "\n",
      "Epoch 300\n",
      "Discriminator loss: -0.0542\n",
      "\n",
      "Epoch 301\n",
      "Discriminator loss: -0.135\n",
      "epsilon = 30.69782064892823, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0155\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7836166924265843\n",
      "Train accuracy: 0.759080370942813\n",
      "Test accuracy: 0.7284679878048781\n",
      "\n",
      "Epoch 302\n",
      "Discriminator loss: -0.0787\n",
      "\n",
      "Epoch 303\n",
      "Discriminator loss: -0.0973\n",
      "change\n",
      "epsilon = 30.86568835606984, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0198\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7832302936630603\n",
      "Train accuracy: 0.7646831530139103\n",
      "Test accuracy: 0.7644817073170732\n",
      "\n",
      "Epoch 304\n",
      "Discriminator loss: -0.0977\n",
      "\n",
      "Epoch 305\n",
      "Discriminator loss: -0.1011\n",
      "change\n",
      "epsilon = 31.03355606321145, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0177\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.785548686244204\n",
      "Train accuracy: 0.7509659969088099\n",
      "Test accuracy: 0.790110518292683\n",
      "\n",
      "Epoch 306\n",
      "Discriminator loss: -0.0725\n",
      "\n",
      "Epoch 307\n",
      "Discriminator loss: -0.1176\n",
      "change\n",
      "epsilon = 31.20142377035306, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0174\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |███████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7845826893353941\n",
      "Train accuracy: 0.758500772797527\n",
      "Test accuracy: 0.7602896341463414\n",
      "\n",
      "Epoch 308\n",
      "Discriminator loss: -0.1133\n",
      "\n",
      "Epoch 309\n",
      "Discriminator loss: -0.0948\n",
      "change\n",
      "epsilon = 31.36929147749467, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0182\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7523183925811437\n",
      "Test accuracy: 0.7883955792682927\n",
      "\n",
      "Epoch 310\n",
      "Discriminator loss: -0.0888\n",
      "\n",
      "Epoch 311\n",
      "Discriminator loss: -0.0902\n",
      "epsilon = 31.53715918463628, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0169\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7840030911901081\n",
      "Train accuracy: 0.7530911901081917\n",
      "Test accuracy: 0.784203506097561\n",
      "\n",
      "Epoch 312\n",
      "Discriminator loss: -0.106\n",
      "\n",
      "Epoch 313\n",
      "Discriminator loss: -0.0806\n",
      "epsilon = 31.70502689177789, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0191\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |█████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7851622874806801\n",
      "Train accuracy: 0.7606259659969088\n",
      "Test accuracy: 0.7518102134146342\n",
      "\n",
      "Epoch 314\n",
      "Discriminator loss: -0.0976\n",
      "\n",
      "Epoch 315\n",
      "Discriminator loss: -0.0632\n",
      "epsilon = 31.8728945989195, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0169\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7828438948995363\n",
      "Train accuracy: 0.7606259659969088\n",
      "Test accuracy: 0.7722942073170732\n",
      "\n",
      "Epoch 316\n",
      "Discriminator loss: -0.1014\n",
      "\n",
      "Epoch 317\n",
      "Discriminator loss: -0.0879\n",
      "change\n",
      "epsilon = 32.04076230606111, delta = 9.518370454978108e-05\n",
      "Wasserstein distance: 0.0148\n",
      "\n",
      "Data is labelled \n",
      "Calculating accuracy scores for AutoML Classifiers ......\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "AutoML progress: |█████████████████████████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\n",
      "Validation accuracy: 0.7826506955177743\n",
      "Train accuracy: 0.7575347758887172\n",
      "Test accuracy: 0.7886814024390244\n",
      "\n",
      "Privacy budget is exhausted ...\n",
      "WGAN training is stopped ...\n",
      "\n",
      "The execution time: 350 min 57.58 sec\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "32.04076230606111"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_WGAN(adult_pr, my_params, './adult_test_stop_criterion')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Plot a side-by-side histogram of category's occurrencies in original and generated datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "original = pd.read_csv('./adult_test_stop_criterion/OriginalFreq.csv')\n",
    "fake_311 = pd.read_csv('./adult_test_stop_criterion/GeneratedFreq_epoch_311.csv')\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Occurrency_Original</th>\n",
       "      <th>Occurrency_Generated</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Category</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>workclass_?</th>\n",
       "      <td>479</td>\n",
       "      <td>246</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>workclass_Federal-gov</th>\n",
       "      <td>367</td>\n",
       "      <td>297</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>workclass_Local-gov</th>\n",
       "      <td>726</td>\n",
       "      <td>1147</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>workclass_Never-worked</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>workclass_Private</th>\n",
       "      <td>7070</td>\n",
       "      <td>6763</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                        Occurrency_Original  Occurrency_Generated\n",
       "Category                                                         \n",
       "workclass_?                             479                   246\n",
       "workclass_Federal-gov                   367                   297\n",
       "workclass_Local-gov                     726                  1147\n",
       "workclass_Never-worked                    1                     0\n",
       "workclass_Private                      7070                  6763"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = original.set_index('Category').join(fake_311.set_index('Category'), \n",
    "                                    lsuffix = '_Original',rsuffix = '_Generated')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Plot a side-by-side histogram of category's occurrencies in original and generated datasets\n",
    "ax = df.plot(kind ='bar', figsize =(30,7))\n",
    "fig = ax.get_figure()\n",
    "fig.savefig(\"./FreqCompare_epoch_87.png\", bbox_inches = 'tight')\n",
    "plt.close(fig)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAooAAAaBCAYAAAAk/K5cAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 504x2160 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "df.plot(kind='barh', figsize=(7,30))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "acc_test = pd.read_csv('./adult_test_stop_criterion/scores/accuracy_test.txt', header =None)\n",
    "acc_train = pd.read_csv('./adult_test_stop_criterion/scores/accuracy_train.txt', header =None)\n",
    "acc_test.columns = ['acc_test']\n",
    "acc_train.columns = ['acc_train']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "epsilon=pd.read_csv('./adult_test_stop_criterion/scores/epsilon.txt',header =None)\n",
    "epsilon.columns=['epsilon']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>acc_test</th>\n",
       "      <th>acc_train</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.497999</td>\n",
       "      <td>0.503091</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.346894</td>\n",
       "      <td>0.501546</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.537062</td>\n",
       "      <td>0.499227</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.444550</td>\n",
       "      <td>0.505796</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.957127</td>\n",
       "      <td>0.418663</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   acc_test  acc_train\n",
       "0  0.497999   0.503091\n",
       "1  0.346894   0.501546\n",
       "2  0.537062   0.499227\n",
       "3  0.444550   0.505796\n",
       "4  0.957127   0.418663"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = acc_test.join(acc_train)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>acc_test</th>\n",
       "      <th>acc_train</th>\n",
       "      <th>epsilon</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.497999</td>\n",
       "      <td>0.503091</td>\n",
       "      <td>0.448825</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.346894</td>\n",
       "      <td>0.501546</td>\n",
       "      <td>0.745073</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.537062</td>\n",
       "      <td>0.499227</td>\n",
       "      <td>0.965266</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.444550</td>\n",
       "      <td>0.505796</td>\n",
       "      <td>1.145925</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.957127</td>\n",
       "      <td>0.418663</td>\n",
       "      <td>1.303305</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   acc_test  acc_train   epsilon\n",
       "0  0.497999   0.503091  0.448825\n",
       "1  0.346894   0.501546  0.745073\n",
       "2  0.537062   0.499227  0.965266\n",
       "3  0.444550   0.505796  1.145925\n",
       "4  0.957127   0.418663  1.303305"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = df.join(epsilon)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "acc_test      0.719976\n",
       "acc_train     0.698068\n",
       "epsilon      17.188748\n",
       "dtype: float64"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAa8AAAEzCAYAAACGz4w5AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nBldlkiHGAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 504x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "df.iloc[:,0:2].plot.box(figsize=(7,5))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# acc_valid = 0.79\n",
    "df['diversity'] = np.abs(0.79-df['acc_train'])\n",
    "df['similarity'] = np.abs(0.79-df['acc_test'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>acc_test</th>\n",
       "      <th>acc_train</th>\n",
       "      <th>epsilon</th>\n",
       "      <th>diversity</th>\n",
       "      <th>similarity</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.497999</td>\n",
       "      <td>0.503091</td>\n",
       "      <td>0.448825</td>\n",
       "      <td>0.286909</td>\n",
       "      <td>0.292001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.346894</td>\n",
       "      <td>0.501546</td>\n",
       "      <td>0.745073</td>\n",
       "      <td>0.288454</td>\n",
       "      <td>0.443106</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.537062</td>\n",
       "      <td>0.499227</td>\n",
       "      <td>0.965266</td>\n",
       "      <td>0.290773</td>\n",
       "      <td>0.252938</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.444550</td>\n",
       "      <td>0.505796</td>\n",
       "      <td>1.145925</td>\n",
       "      <td>0.284204</td>\n",
       "      <td>0.345450</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.957127</td>\n",
       "      <td>0.418663</td>\n",
       "      <td>1.303305</td>\n",
       "      <td>0.371337</td>\n",
       "      <td>0.167127</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   acc_test  acc_train   epsilon  diversity  similarity\n",
       "0  0.497999   0.503091  0.448825   0.286909    0.292001\n",
       "1  0.346894   0.501546  0.745073   0.288454    0.443106\n",
       "2  0.537062   0.499227  0.965266   0.290773    0.252938\n",
       "3  0.444550   0.505796  1.145925   0.284204    0.345450\n",
       "4  0.957127   0.418663  1.303305   0.371337    0.167127"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD8CAYAAACMwORRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nZmYZ5ORuZpZBTu5mZhn0/wEZuBWvavyB9wAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.iloc[:,3:5].plot.box()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmQAAAGtCAYAAAC4HmhdAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ndt1c5gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.iloc[:,3:5].plot(figsize=(10,7))\n",
    "plt.ylabel('Diversity/Similarity')\n",
    "plt.xlabel('Epochs')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ideally, both measures,i.e. divesrity and similarity has to be = 0. The good ratio quality/utility of generated data was achieved after 50-60 epochs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}