test.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def general_analysis(file):\n",
    "    #Create Pandas Series\n",
    "    f = pd.read_csv(file)\n",
    "    \n",
    "    #add dummy columns for thedevice and themodule (if applicable) to add up.\n",
    "    f = f.join(pd.get_dummies(f['thedevice']))\n",
    "    if 'themodule' in f.columns.values:\n",
    "        f = f.join(pd.get_dummies(f['themodule']))\n",
    "        \n",
    "    return f"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def create_col(file, colname):\n",
    "    file[colname] = pd.DataFrame([0 for x in range(len(file.index))])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def sums(file):\n",
    "    return file.sum()\n",
    "\n",
    "def shape(file):\n",
    "    return file.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Computes Pearson correlation between two columns of a file.\n",
    "def correlation(file, col1, col2):\n",
    "    return file[col1].corr(file[col2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def wl_test_analysis(f):\n",
    "    '''\n",
    "    Add createnormal to check if user created account w/o social media\n",
    "    Add sm to check if user created account with Facebook or Google.\n",
    "    Add falselogin for when user probably clicked on social media option, but \n",
    "    did not follow through.\n",
    "    This is impossible to check for non-sm option, since false login and never trying to\n",
    "    create an account both result in 0.\n",
    "    Do search for created accounts with F, G, and false logins to be device specific as well.\n",
    "    '''\n",
    "    create_col(f, 'C')\n",
    "    create_col(f, 'L')\n",
    "    create_col(f, 'signin_count')\n",
    "    create_col(f, 'desktopcreate')\n",
    "    create_col(f, 'mobilecreate')\n",
    "    create_col(f, 'tabletcreate')\n",
    "    create_col(f, 'normalcreate')\n",
    "    create_col(f, 'sm')\n",
    "    create_col(f, 'falselogin')\n",
    "    create_col(f, 'desktopg')\n",
    "    create_col(f, 'desktopf')\n",
    "    create_col(f, 'desktopnormal')\n",
    "    create_col(f, 'desktop_false')\n",
    "    create_col(f, 'mobileg')\n",
    "    create_col(f, 'mobilef')\n",
    "    create_col(f, 'mobilenormal')\n",
    "    create_col(f, 'mobile_false')\n",
    "    create_col(f, 'tabletg')\n",
    "    create_col(f, 'tabletf')\n",
    "    create_col(f, 'tabletnormal')\n",
    "    create_col(f, 'tablet_false')\n",
    "    \n",
    "    for i in range(len(f.index)):\n",
    "        letters = f.at[i, 'login_or_create'].split(',')\n",
    "        \n",
    "        if 'C' in letters:\n",
    "            f.at[i, 'C'] += 1\n",
    "            if f.at[i, 'themodule'] == 'F' or f.at[i, 'themodule'] == 'G':\n",
    "                f.at[i, 'sm'] += 1 \n",
    "                \n",
    "            if f.at[i, '0'] == 1:\n",
    "                f.at[i, 'normalcreate'] += 1\n",
    "\n",
    "            if f.at[i, 'DESKTOP'] == 1:\n",
    "                f.at[i, 'desktopcreate'] += 1\n",
    "                if f.at[i, 'themodule'] == 'F':\n",
    "                    f.at[i, 'desktopf'] += 1 \n",
    "                elif f.at[i, 'themodule'] == 'G':\n",
    "                    f.at[i, 'desktopg'] += 1 \n",
    "                elif f.at[i, 'themodule'] == '0':\n",
    "                    f.at[i, 'desktopnormal'] += 1\n",
    "                    \n",
    "            elif f.at[i, 'MOBILE'] == 1:\n",
    "                f.at[i, 'mobilecreate'] += 1 \n",
    "                if f.at[i, 'themodule'] == 'F':\n",
    "                    f.at[i, 'mobilef'] += 1 \n",
    "                elif f.at[i, 'themodule'] == 'G':\n",
    "                    f.at[i, 'mobileg'] += 1 \n",
    "                elif f.at[i, 'themodule'] == '0':\n",
    "                    f.at[i, 'mobilenormal'] += 1\n",
    "                \n",
    "            elif f.at[i, 'TABLET'] == 1:\n",
    "                f.at[i, 'tabletcreate'] += 1\n",
    "                if f.at[i, 'themodule'] == 'F':\n",
    "                    f.at[i, 'tabletf'] += 1 \n",
    "                elif f.at[i, 'themodule'] == 'G':\n",
    "                    f.at[i, 'tabletg'] += 1 \n",
    "                elif f.at[i, 'themodule'] == '0':\n",
    "                    f.at[i, 'tabletnormal'] += 1\n",
    "                    \n",
    "        if 'L' in letters:\n",
    "            f.at[i, 'L'] += 1\n",
    "        for letter in letters:\n",
    "            if letter == 'L':\n",
    "                f.at[i, 'signin_count'] += 1\n",
    "                    \n",
    "        if 'C' not in letters and 'L' not in letters:\n",
    "            if f.at[i, 'themodule'] == 'F' or f.at[i, 'themodule'] == 'G':\n",
    "                f.at[i, 'falselogin'] += 1\n",
    "                if f.at[i, 'DESKTOP'] == 1:\n",
    "                    f.at[i, 'desktop_false'] += 1\n",
    "                if f.at[i, 'TABLET'] == 1:\n",
    "                    f.at[i, 'tablet_false'] += 1\n",
    "                if f.at[i, 'MOBILE'] == 1:\n",
    "                    f.at[i, 'mobile_false'] += 1\n",
    "        \n",
    "        user_check = f.at[i, 'isloggedin_r'].split(',')\n",
    "        for event in user_check:\n",
    "            if event == 'user':\n",
    "                f.at[i, 'signin_count'] += 1 \n",
    "    \n",
    "    # Return data as a Pandas Series, inclusive of all dummy columns.\n",
    "    return f"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def co_test_analysis(f):\n",
    "    # Same notes and labeling as wl_test_analysis\n",
    "    create_col(f, 'sm')\n",
    "    create_col(f, 'normalcreate')\n",
    "    # Still social media specific, impossible to tell with normal login vs no login.\n",
    "    create_col(f, 'falselogin')\n",
    "    create_col(f, 'desktopcreate')\n",
    "    create_col(f, 'desktopg')\n",
    "    create_col(f, 'desktopf')\n",
    "    create_col(f, 'mobilecreate')\n",
    "    create_col(f, 'mobileg')\n",
    "    create_col(f, 'mobilef')\n",
    "    create_col(f, 'tabletcreate')\n",
    "    create_col(f, 'tabletg')\n",
    "    create_col(f, 'tabletf')\n",
    "    \n",
    "    for i in range(len(f.index)):\n",
    "        if (f.at[i, 'themodule'] == 'F' or f.at[i, 'themodule'] == 'G') and f.at[i, 'reg_success'] == 1:\n",
    "            f.at[i, 'sm'] += 1 \n",
    "        elif (f.at[i, 'themodule'] == 'F' or f.at[i, 'themodule'] == 'G') and f.at[i, 'reg_success'] == 0:\n",
    "            f.at[i, 'falselogin'] += 1 \n",
    "        if (f.at[i, 'themodule'] == '0') and f.at[i, 'reg_success'] == 1:\n",
    "            f.at[i, 'normalcreate'] += 1 \n",
    "        \n",
    "        if f.at[i, 'reg_success'] == 1 and f.at[i, 'DESKTOP'] == 1:\n",
    "            f.at[i, 'desktopcreate'] += 1\n",
    "        elif f.at[i, 'reg_success'] == 1 and f.at[i, 'MOBILE'] == 1:\n",
    "            f.at[i, 'mobilecreate'] += 1\n",
    "        elif f.at[i, 'reg_success'] == 1 and f.at[i, 'TABLET'] == 1:\n",
    "            f.at[i, 'tabletcreate'] += 1\n",
    "        if f.at[i, 'themodule'] == 'F' and f.at[i, 'DESKTOP'] == 1:\n",
    "            f.at[i, 'desktopf'] += 1 \n",
    "        elif f.at[i, 'themodule'] == 'G' and f.at[i, 'DESKTOP'] == 1:\n",
    "            f.at[i, 'desktopg'] += 1 \n",
    "        elif f.at[i, 'themodule'] == 'F' and f.at[i, 'MOBILE'] == 1:\n",
    "            f.at[i, 'mobilef'] += 1 \n",
    "        elif f.at[i, 'themodule'] == 'G' and f.at[i, 'MOBILE'] == 1:\n",
    "            f.at[i, 'mobileg'] += 1 \n",
    "        elif f.at[i, 'themodule'] == 'F' and f.at[i, 'TABLET'] == 1:\n",
    "            f.at[i, 'tabletf'] += 1 \n",
    "        elif f.at[i, 'themodule'] == 'G' and f.at[i, 'TABLET'] == 1:\n",
    "            f.at[i, 'tabletg'] += 1 \n",
    "\n",
    "            \n",
    "    return f"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def wl_general_analysis(f):\n",
    "    '''\n",
    "    Add column C to check if the user created account\n",
    "    Add column L to check how many time the user logged in\n",
    "    Add column signin_count to see how many events involved logging in or already being signed in\n",
    "    Though some of these exact columns were created in in wl_test_analysis, I did not want to run\n",
    "    the for loop an extra time by running this function on the test csv. \n",
    "    Same for co_general_analysis.\n",
    "    '''\n",
    "    create_col(f, 'C')\n",
    "    create_col(f, 'L')\n",
    "    create_col(f, 'signin_count')\n",
    "    create_col(f, 'desktopcreate')\n",
    "    create_col(f, 'mobilecreate')\n",
    "    create_col(f, 'tabletcreate')\n",
    "    \n",
    "    for i in range(len(f.index)):\n",
    "        letters = f.at[i, 'login_or_create'].split(',')\n",
    "        if 'C' in letters:\n",
    "            f.at[i, 'C'] += 1\n",
    "            \n",
    "            if f.at[i, 'DESKTOP'] == 1:\n",
    "                f.at[i, 'desktopcreate'] += 1 \n",
    "            elif f.at[i, 'MOBILE'] == 1:\n",
    "                f.at[i, 'mobilecreate'] += 1 \n",
    "            elif f.at[i, 'TABLET'] == 1:\n",
    "                f.at[i, 'tabletcreate'] += 1\n",
    "                \n",
    "        if 'L' in letters:\n",
    "            f.at[i, 'L'] += 1\n",
    "        for letter in letters:\n",
    "            if letter == 'L':\n",
    "                f.at[i, 'signin_count'] += 1\n",
    "                    \n",
    "        user_check = f.at[i, 'isloggedin_r'].split(',')\n",
    "        for event in user_check:\n",
    "            if event == 'user':\n",
    "                f.at[i, 'signin_count'] += 1\n",
    "                    \n",
    "    return f"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def co_general_analysis(f):\n",
    "    # Same notes and labeling as wl_general_analysis\n",
    "    create_col(f, 'desktopcreate')\n",
    "    create_col(f, 'mobilecreate')\n",
    "    create_col(f, 'tabletcreate')\n",
    "    \n",
    "    for i in range(len(f.index)):\n",
    "        if f.at[i, 'reg_success'] == 1 and f.at[i, 'DESKTOP'] == 1:\n",
    "            f.at[i, 'desktopcreate'] += 1\n",
    "        elif f.at[i, 'reg_success'] == 1 and f.at[i, 'MOBILE'] == 1:\n",
    "            f.at[i, 'mobilecreate'] += 1\n",
    "        elif f.at[i, 'reg_success'] == 1 and f.at[i, 'TABLET'] == 1:\n",
    "            f.at[i, 'tabletcreate'] += 1\n",
    "    \n",
    "    return f"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "eventcount       5816\n",
      "reg_success       446\n",
      "DESKTOP          2296\n",
      "MOBILE            695\n",
      "TABLET            352\n",
      "desktopcreate     361\n",
      "mobilecreate       45\n",
      "tabletcreate       40\n",
      "dtype: float64\n",
      "(3345, 12)\n"
     ]
    }
   ],
   "source": [
    "f = general_analysis('after_co.csv')\n",
    "co_after = co_general_analysis(f)\n",
    "print(sums(co_after))\n",
    "# Number of users in test is first part of tuple.\n",
    "print(shape(co_after))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "eventcount       24412\n",
      "reg_success       1921\n",
      "DESKTOP           8978\n",
      "MOBILE            2897\n",
      "TABLET            1613\n",
      "desktopcreate     1520\n",
      "mobilecreate       200\n",
      "tabletcreate       201\n",
      "dtype: float64\n",
      "(13491, 12)\n"
     ]
    }
   ],
   "source": [
    "f = general_analysis('before_co.csv')\n",
    "co_before = co_general_analysis(f)\n",
    "print(sums(co_before))\n",
    "# Number of users in test is first part of tuple.\n",
    "print(shape(co_before))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "eventcount       15328\n",
      "DESKTOP           1686\n",
      "MOBILE            1073\n",
      "TABLET             424\n",
      "0                 2724\n",
      "F                  329\n",
      "G                  134\n",
      "C                  832\n",
      "L                  761\n",
      "signin_count      2987\n",
      "desktopcreate      470\n",
      "mobilecreate       265\n",
      "tabletcreate        97\n",
      "normalcreate       579\n",
      "sm                 253\n",
      "falselogin         143\n",
      "desktopg            57\n",
      "desktopf            95\n",
      "desktopnormal      318\n",
      "desktop_false       53\n",
      "mobileg             19\n",
      "mobilef             49\n",
      "mobilenormal       197\n",
      "mobile_false        71\n",
      "tabletg             17\n",
      "tabletf             16\n",
      "tabletnormal        64\n",
      "tablet_false        19\n",
      "dtype: float64\n",
      "(3187, 36)\n"
     ]
    }
   ],
   "source": [
    "f = general_analysis('test_wl.csv')\n",
    "wl_test = wl_test_analysis(f)\n",
    "print(sums(wl_test))\n",
    "# Number of users in test is first part of tuple.\n",
    "print(shape(wl_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "eventcount       24067\n",
      "reg_success       1040\n",
      "DESKTOP           9875\n",
      "MOBILE            2765\n",
      "TABLET            1775\n",
      "0                14236\n",
      "F                  124\n",
      "G                   67\n",
      "sm                 140\n",
      "normalcreate       900\n",
      "falselogin          51\n",
      "desktopcreate      795\n",
      "desktopg            47\n",
      "desktopf            82\n",
      "mobilecreate       133\n",
      "mobileg             12\n",
      "mobilef             26\n",
      "tabletcreate       112\n",
      "tabletg              8\n",
      "tabletf             16\n",
      "dtype: float64\n",
      "(14427, 25)\n"
     ]
    }
   ],
   "source": [
    "f = general_analysis('test_co.csv')\n",
    "co_test = co_test_analysis(f)\n",
    "print(sums(co_test))\n",
    "# Number of users in test is first part of tuple.\n",
    "print(shape(co_test))\n",
    "# Recognize that this could be bias since we don't know of the reg_success that's not SM \n",
    "# if they already had an account before test or not."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "eventcount       3559\n",
      "DESKTOP           429\n",
      "MOBILE            233\n",
      "TABLET             92\n",
      "C                 177\n",
      "L                 212\n",
      "signin_count      737\n",
      "desktopcreate     105\n",
      "mobilecreate       50\n",
      "tabletcreate       22\n",
      "dtype: float64\n",
      "(756, 17)\n"
     ]
    }
   ],
   "source": [
    "f = general_analysis('after_wl.csv')\n",
    "wl_after = wl_general_analysis(f)\n",
    "print(sums(wl_after))\n",
    "# Number of users in test is first part of tuple.\n",
    "print(shape(wl_after))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_time              2016-02-28T23:32:17.809-05002016-02-28T23:26:5...\n",
      "eventcount                                                     15353\n",
      "firstpage          wishwishwishwishwishwishwishwishwishwishwishwi...\n",
      "theugid            15765b3d1532b34b8456d6015765b3d1532b34b845740f...\n",
      "thedevice          DESKTOPDESKTOPTABLETDESKTOPDESKTOPTABLETMOBILE...\n",
      "isloggedin_r       user by-session,user by-session,user by-sessio...\n",
      "time_hhmmss_r      23:32:17,23:32:17,23:32:1823:26:52,23:26:52,23...\n",
      "login_or_create    N,N,NN,N,NN,N,NN,N,NN,N,N,C,N,N,N,NN,N,NN,N,N,...\n",
      "DESKTOP                                                         1825\n",
      "MOBILE                                                          1120\n",
      "TABLET                                                           377\n",
      "C                                                                877\n",
      "L                                                                812\n",
      "signin_count                                                    3057\n",
      "desktopcreate                                                    521\n",
      "mobilecreate                                                     274\n",
      "tabletcreate                                                      82\n",
      "dtype: object\n",
      "(3322, 17)\n"
     ]
    }
   ],
   "source": [
    "# Not sure why exact same functions as above are providing extra data here.\n",
    "f = general_analysis('before_wl.csv')\n",
    "wl_before = wl_general_analysis(f)\n",
    "print(sums(wl_before))\n",
    "# Number of users in test is first part of tuple.\n",
    "print(shape(wl_before))"
   ]
  }
 ],
 "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": "3.5.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}