classifier-evaluation-using-confusion-matrix.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "b5b632c2-8ac7-4aaa-8ad5-30ed5b4b90ef",
    "_execution_state": "idle",
    "_uuid": "bd8c1f68153cc5578e790b07fb45e9cac46ea397"
   },
   "source": [
    "\n",
    "#Evaluation BEFORE and AFTER building a Machine Learning model##\n",
    "\n",
    "This basic Notebook explores methods for evaluation BEFORE and AFTER building a model. Evaluation of data and evaluation of model performance are key to any successful Machine Learning implementation so hope you find these tips helpful. This notebook was authored by Diego Schapira."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "_cell_guid": "95b75d05-9d30-416b-bdb6-d6638bcf88a5",
    "_execution_state": "idle",
    "_uuid": "248fc410977b4b538a565ce6a0a7a773a41a1045"
   },
   "outputs": [],
   "source": [
    "# First of all let's import basic libraries and data to get started\n",
    "import numpy as np\n",
    "from numpy  import array\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "iris = pd.read_csv(\"data/iris-species/iris.csv\") # load data\n",
    "iris.drop('Id',axis=1,inplace=True) # Id column is redundant as df already assigns index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "_cell_guid": "ca32518c-9c2a-4609-8b65-7208fda9ce1f",
    "_execution_state": "idle",
    "_uuid": "57d28965cf16828455349369fe8297fb201d7cef"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 150 entries, 0 to 149\n",
      "Data columns (total 5 columns):\n",
      "SepalLengthCm    150 non-null float64\n",
      "SepalWidthCm     150 non-null float64\n",
      "PetalLengthCm    150 non-null float64\n",
      "PetalWidthCm     150 non-null float64\n",
      "Species          150 non-null object\n",
      "dtypes: float64(4), object(1)\n",
      "memory usage: 5.9+ KB\n"
     ]
    }
   ],
   "source": [
    "# DATA EXPLORATION...\n",
    "\n",
    "# Before starting pre-processing, basic descriptive statistics can help identify if \n",
    "# scaling/normalization of data is needed, or how variance varies between \n",
    "# features which is useful information for feature selection or at least understanding which \n",
    "# features might be most important for classification\n",
    "\n",
    "# dataframe.info gives basic information on data integrity (data types and detection of NaN values)\n",
    "iris.info()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "_cell_guid": "c0a194e0-a8d0-4cef-9aca-fd9475c1e262",
    "_execution_state": "idle",
    "_uuid": "dd3f74fb488c6050d302ef6bc1fe714a03079a02"
   },
   "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>SepalLengthCm</th>\n",
       "      <th>SepalWidthCm</th>\n",
       "      <th>PetalLengthCm</th>\n",
       "      <th>PetalWidthCm</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>150.000000</td>\n",
       "      <td>150.000000</td>\n",
       "      <td>150.000000</td>\n",
       "      <td>150.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>5.843333</td>\n",
       "      <td>3.054000</td>\n",
       "      <td>3.758667</td>\n",
       "      <td>1.198667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.828066</td>\n",
       "      <td>0.433594</td>\n",
       "      <td>1.764420</td>\n",
       "      <td>0.763161</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>4.300000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.100000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>5.100000</td>\n",
       "      <td>2.800000</td>\n",
       "      <td>1.600000</td>\n",
       "      <td>0.300000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>5.800000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>4.350000</td>\n",
       "      <td>1.300000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>6.400000</td>\n",
       "      <td>3.300000</td>\n",
       "      <td>5.100000</td>\n",
       "      <td>1.800000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>7.900000</td>\n",
       "      <td>4.400000</td>\n",
       "      <td>6.900000</td>\n",
       "      <td>2.500000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       SepalLengthCm  SepalWidthCm  PetalLengthCm  PetalWidthCm\n",
       "count     150.000000    150.000000     150.000000    150.000000\n",
       "mean        5.843333      3.054000       3.758667      1.198667\n",
       "std         0.828066      0.433594       1.764420      0.763161\n",
       "min         4.300000      2.000000       1.000000      0.100000\n",
       "25%         5.100000      2.800000       1.600000      0.300000\n",
       "50%         5.800000      3.000000       4.350000      1.300000\n",
       "75%         6.400000      3.300000       5.100000      1.800000\n",
       "max         7.900000      4.400000       6.900000      2.500000"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Next I'm using dataframe.describe function. Works for both object and numeric\n",
    "iris.describe()\n",
    "\n",
    "# Means are in the same order of magnitude for all features so scaling might not be beneficial.\n",
    "# If mean values were of different orders of magnitude, scaling could significantly \n",
    "# improve accuracy of a classifier."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "_cell_guid": "d5ec14dd-3973-4ff6-8acd-b30fee51e08b",
    "_execution_state": "idle",
    "_uuid": "2ee461305110eeb4e4de96fcd1e147ffda009219"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_scaled type is <class 'numpy.ndarray'>\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x504 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "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>s_SepalLength</th>\n",
       "      <th>s_SepalWidth</th>\n",
       "      <th>s_PetalLength</th>\n",
       "      <th>s_PetalWidth</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>150.000000</td>\n",
       "      <td>150.000000</td>\n",
       "      <td>150.000000</td>\n",
       "      <td>150.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.428704</td>\n",
       "      <td>0.439167</td>\n",
       "      <td>0.467571</td>\n",
       "      <td>0.457778</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.230018</td>\n",
       "      <td>0.180664</td>\n",
       "      <td>0.299054</td>\n",
       "      <td>0.317984</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.222222</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.101695</td>\n",
       "      <td>0.083333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.416667</td>\n",
       "      <td>0.416667</td>\n",
       "      <td>0.567797</td>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.583333</td>\n",
       "      <td>0.541667</td>\n",
       "      <td>0.694915</td>\n",
       "      <td>0.708333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       s_SepalLength  s_SepalWidth  s_PetalLength  s_PetalWidth\n",
       "count     150.000000    150.000000     150.000000    150.000000\n",
       "mean        0.428704      0.439167       0.467571      0.457778\n",
       "std         0.230018      0.180664       0.299054      0.317984\n",
       "min         0.000000      0.000000       0.000000      0.000000\n",
       "25%         0.222222      0.333333       0.101695      0.083333\n",
       "50%         0.416667      0.416667       0.567797      0.500000\n",
       "75%         0.583333      0.541667       0.694915      0.708333\n",
       "max         1.000000      1.000000       1.000000      1.000000"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Although not essential for this dataset let's see if scaling provides additional insights.\n",
    "\n",
    "# First we need to split X, Y into separate sets to feed X to the scaler.\n",
    "X = iris.drop('Species',1)\n",
    "Y = iris.Species\n",
    "\n",
    "# Scaling of X\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "scaler = MinMaxScaler()\n",
    "X_scaled = scaler.fit_transform(X)\n",
    "print('X_scaled type is',type(X_scaled))\n",
    "\n",
    "# As output of scaler is np array I'll transform back to df for easier exploration and plotting\n",
    "X_scaled_df = pd.DataFrame(X_scaled,columns=['s_SepalLength','s_SepalWidth',\n",
    "                                             's_PetalLength','s_PetalWidth'])\n",
    "df = pd.concat([X_scaled_df,Y],axis=1)\n",
    "\n",
    "# Notice x-axis on subplots are all the same for all features (0 to 1) after scaling.\n",
    "fig = plt.figure(figsize=(12,7))\n",
    "fig.suptitle('Frequency Distribution of Features by Species ',fontsize=20)\n",
    "\n",
    "ax1 = fig.add_subplot(221)\n",
    "df.groupby(\"Species\").s_PetalLength.plot(kind='hist',alpha=0.8,legend=True,title='s_PetalLength')\n",
    "ax2 = fig.add_subplot(222,sharey=ax1)\n",
    "df.groupby(\"Species\").s_PetalWidth.plot(kind='hist',alpha=0.8,legend=True,title='s_PetalWidth')\n",
    "ax3 = fig.add_subplot(223,sharey=ax1)\n",
    "df.groupby(\"Species\").s_SepalLength.plot(kind='hist',alpha=0.8,legend=True,title='s_SepalLength')\n",
    "ax4 = fig.add_subplot(224,sharey=ax1)\n",
    "df.groupby(\"Species\").s_SepalWidth.plot(kind='hist',alpha=0.8,legend=True,title='s_SepalWidth')\n",
    "\n",
    "plt.tight_layout(pad=4, w_pad=1, h_pad=1.5)\n",
    "plt.show()\n",
    "X_scaled_df.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "_cell_guid": "043b07d2-d713-4037-a127-5c7fdc77c930",
    "_execution_state": "idle",
    "_uuid": "cb5411e30b6d8b3e221b0af4e022aba540daca69"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/pantelis/anaconda3/lib/python3.6/site-packages/seaborn/axisgrid.py:230: UserWarning: The `size` paramter has been renamed to `height`; please update your code.\n",
      "  warnings.warn(msg, UserWarning)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 535.5x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Another useful visualization I like to use during exploration is a 2D PCA plot.\n",
    "# It can quickly indicate how easy or difficult the classification problem is.\n",
    "# This is particularly relevant for high-dimensional datasets.\n",
    "\n",
    "from sklearn.decomposition import PCA\n",
    "# spliting input and target\n",
    "X = iris.drop('Species',1)\n",
    "Y = iris.Species\n",
    "\n",
    "pca = PCA(n_components=2).fit_transform(X) # pca output is an array\n",
    "pca_df = pd.DataFrame(pca,columns=['pca_1','pca_2']) # transforming back to df\n",
    "pca_Y = pd.concat([pca_df, Y],axis=1)\n",
    "\n",
    "# The 3 species cluster nicely which is a good indication a classifier can be trained \n",
    "# at high accuracy. \n",
    "# It is also expected that accuracy of identifying setosa will be higher.\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "sns.FacetGrid(pca_Y, hue=\"Species\", palette=\"Set1\", size=6).map(plt.scatter, \"pca_1\", \"pca_2\").add_legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "_cell_guid": "53fc1bb3-142e-4167-87d3-c423249272af",
    "_execution_state": "idle",
    "_uuid": "fcecc1e77516ceb1c372a29cbe1f7c180dd2b8a6"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train sample size (105, 4) <class 'pandas.core.frame.DataFrame'>\n",
      "test sample size (45, 4) <class 'pandas.core.frame.DataFrame'>\n"
     ]
    }
   ],
   "source": [
    "# Let's now build a classifier and evaluate accuracy\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "x_train, x_test, y_train, y_test = train_test_split(X, Y, test_size=0.3, random_state = 0)\n",
    "print(\"train sample size\",x_train.shape,type(x_train))\n",
    "print(\"test sample size\",x_test.shape,type(x_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "_cell_guid": "ebce6a7e-f8de-4ac3-8c4a-4388b662891c",
    "_execution_state": "idle",
    "_uuid": "d2e383b2870d0af9ab5cdb73f5ed5af4a3323985"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 396x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.svm import SVC\n",
    "from sklearn.metrics import accuracy_score, confusion_matrix, precision_recall_fscore_support\n",
    "\n",
    "clf = SVC(kernel = 'linear').fit(x_train,y_train)\n",
    "clf.predict(x_train)\n",
    "y_pred = clf.predict(x_test)\n",
    "\n",
    "# Creates a confusion matrix\n",
    "cm = confusion_matrix(y_test, y_pred) \n",
    "\n",
    "# Transform to df for easier plotting\n",
    "cm_df = pd.DataFrame(cm,\n",
    "                     index = ['setosa','versicolor','virginica'], \n",
    "                     columns = ['setosa','versicolor','virginica'])\n",
    "\n",
    "plt.figure(figsize=(5.5,4))\n",
    "sns.heatmap(cm_df, annot=True)\n",
    "plt.title('SVM Linear Kernel \\nAccuracy:{0:.3f}'.format(accuracy_score(y_test, y_pred)))\n",
    "plt.ylabel('True label')\n",
    "plt.xlabel('Predicted label')\n",
    "plt.show()\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "e4c5b7a9-68da-4b09-bd7e-1b4690a2bf6f",
    "_execution_state": "idle",
    "_uuid": "294fa25aae5511672f1b7da3c9a82ac69f24c461"
   },
   "source": [
    "In this example the model overall accuracy on the test set was 0.978 (Accurate predictions/All or True Positives/All).\n",
    " \n",
    "However, the confusion matrix gives additional insight into accuracy by class and intuition for precision and recall efficiency.\n",
    " \n",
    "An insight we can get from the matrix is that the model was very accurate at classifying setosa and versicolor (True Positive/All = 1.0). However, accuracy for virginica was lower (11/12 = 0.917).\n",
    "\n",
    "If for any reason, successful classification of virginica was particularly desired to the use case (e.g. it's a poisonous plant), then the confusion matrix will help highlight differences between the classes. \n",
    "\n",
    "sklearn.metrics module has many functions that calculate evaluation scores like precision and recall. In this example I wanted to show the intuition for Machine Learning model evaluation beyond a simple assessment of overall test/pred accuracy."
   ]
  }
 ],
 "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": "3.6.7"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}