Sparkify.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sparkify Project Workspace\n",
    "This workspace contains a tiny subset (128MB) of the full dataset available (12GB). Feel free to use this workspace to build your project, or to explore a smaller subset with Spark before deploying your cluster on the cloud. Instructions for setting up your Spark cluster is included in the last lesson of the Extracurricular Spark Course content.\n",
    "\n",
    "You can follow the steps below to guide your data analysis and model building portion of this project."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import libraries\n",
    "from pyspark.sql import SparkSession\n",
    "from pyspark.sql.functions import avg, col, concat, desc, explode, lit, min, max, split, udf, isnull\n",
    "from pyspark.sql.types import IntegerType\n",
    "from pyspark.sql import Window\n",
    "\n",
    "from pyspark.ml import Pipeline\n",
    "from pyspark.ml.classification import LogisticRegression, RandomForestClassifier,LinearSVC, GBTClassifier, DecisionTreeClassifier, NaiveBayes\n",
    "from pyspark.ml.evaluation import MulticlassClassificationEvaluator\n",
    "from pyspark.ml.feature import CountVectorizer, IDF, Normalizer, PCA, RegexTokenizer, StandardScaler, StopWordsRemover, StringIndexer, VectorAssembler\n",
    "from pyspark.ml.regression import LinearRegression\n",
    "from pyspark.ml.tuning import CrossValidator, ParamGridBuilder\n",
    "\n",
    "import re\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "import datetime\n",
    "from time import time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# create a Spark session\n",
    "spark = SparkSession \\\n",
    "    .builder \\\n",
    "    .appName(\"Sparkify\") \\\n",
    "    .getOrCreate()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load and Clean Dataset\n",
    "In this workspace, the mini-dataset file is `mini_sparkify_event_data.json`. Load and clean the dataset, checking for invalid or missing data - for example, records without userids or sessionids. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Preprocessing Steps\n",
    "\n",
    "I will perform below steps in the preprocessing part to discover and analyze the dataset in a better way. \n",
    "\n",
    "1. Loading the data from json file and make transformation to pandas dataframe.\n",
    "2. Dealing with missing variables.\n",
    "    1. Analyzing the variables of columns and removing empty entries, '', in userID.   \n",
    "    2. Identifying the missing features and removing or replacing them.\n",
    "3. Converting the timestamp column in more readible way. \n",
    "4. Converting gender information to binary variable from Female to Male. \n",
    "\n",
    "After performing these steps, *Churn* will be defined according to Cancellation Confirmation feature. In the Churn column, 1 will represent the user has churned and 0 otherwise. \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Loading The Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- artist: string (nullable = true)\n",
      " |-- auth: string (nullable = true)\n",
      " |-- firstName: string (nullable = true)\n",
      " |-- gender: string (nullable = true)\n",
      " |-- itemInSession: long (nullable = true)\n",
      " |-- lastName: string (nullable = true)\n",
      " |-- length: double (nullable = true)\n",
      " |-- level: string (nullable = true)\n",
      " |-- location: string (nullable = true)\n",
      " |-- method: string (nullable = true)\n",
      " |-- page: string (nullable = true)\n",
      " |-- registration: long (nullable = true)\n",
      " |-- sessionId: long (nullable = true)\n",
      " |-- song: string (nullable = true)\n",
      " |-- status: long (nullable = true)\n",
      " |-- ts: long (nullable = true)\n",
      " |-- userAgent: string (nullable = true)\n",
      " |-- userId: string (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "data_path = 'mini_sparkify_event_data.json'\n",
    "df = spark.read.json(data_path)\n",
    "# df.persist()\n",
    "# See the frame schema\n",
    "df.printSchema()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Since I am more familiar with pandas dataframe, I will convert the spark object to pandas df.\n",
    "\n",
    "[1] https://stackoverflow.com/questions/29226210/what-is-the-spark-dataframe-method-topandas-actually-doing\n",
    "\n",
    "[2] https://www.oreilly.com/library/view/pyspark-cookbook/9781788835367/fe13b699-9b45-4295-a0cb-6375fa98a3e3.xhtml"
   ]
  },
  {
   "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>artist</th>\n",
       "      <th>auth</th>\n",
       "      <th>firstName</th>\n",
       "      <th>gender</th>\n",
       "      <th>itemInSession</th>\n",
       "      <th>lastName</th>\n",
       "      <th>length</th>\n",
       "      <th>level</th>\n",
       "      <th>location</th>\n",
       "      <th>method</th>\n",
       "      <th>page</th>\n",
       "      <th>registration</th>\n",
       "      <th>sessionId</th>\n",
       "      <th>song</th>\n",
       "      <th>status</th>\n",
       "      <th>ts</th>\n",
       "      <th>userAgent</th>\n",
       "      <th>userId</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Martha Tilston</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Colin</td>\n",
       "      <td>M</td>\n",
       "      <td>50</td>\n",
       "      <td>Freeman</td>\n",
       "      <td>277.89016</td>\n",
       "      <td>paid</td>\n",
       "      <td>Bakersfield, CA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1.538173e+12</td>\n",
       "      <td>29</td>\n",
       "      <td>Rockpools</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352117000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Five Iron Frenzy</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>79</td>\n",
       "      <td>Long</td>\n",
       "      <td>236.09424</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1.538332e+12</td>\n",
       "      <td>8</td>\n",
       "      <td>Canada</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352180000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Adam Lambert</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Colin</td>\n",
       "      <td>M</td>\n",
       "      <td>51</td>\n",
       "      <td>Freeman</td>\n",
       "      <td>282.82730</td>\n",
       "      <td>paid</td>\n",
       "      <td>Bakersfield, CA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1.538173e+12</td>\n",
       "      <td>29</td>\n",
       "      <td>Time For Miracles</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352394000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Enigma</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>80</td>\n",
       "      <td>Long</td>\n",
       "      <td>262.71302</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1.538332e+12</td>\n",
       "      <td>8</td>\n",
       "      <td>Knocking On Forbidden Doors</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352416000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Daft Punk</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Colin</td>\n",
       "      <td>M</td>\n",
       "      <td>52</td>\n",
       "      <td>Freeman</td>\n",
       "      <td>223.60771</td>\n",
       "      <td>paid</td>\n",
       "      <td>Bakersfield, CA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1.538173e+12</td>\n",
       "      <td>29</td>\n",
       "      <td>Harder Better Faster Stronger</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352676000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             artist       auth firstName gender  itemInSession lastName  \\\n",
       "0    Martha Tilston  Logged In     Colin      M             50  Freeman   \n",
       "1  Five Iron Frenzy  Logged In     Micah      M             79     Long   \n",
       "2      Adam Lambert  Logged In     Colin      M             51  Freeman   \n",
       "3            Enigma  Logged In     Micah      M             80     Long   \n",
       "4         Daft Punk  Logged In     Colin      M             52  Freeman   \n",
       "\n",
       "      length level                        location method      page  \\\n",
       "0  277.89016  paid                 Bakersfield, CA    PUT  NextSong   \n",
       "1  236.09424  free  Boston-Cambridge-Newton, MA-NH    PUT  NextSong   \n",
       "2  282.82730  paid                 Bakersfield, CA    PUT  NextSong   \n",
       "3  262.71302  free  Boston-Cambridge-Newton, MA-NH    PUT  NextSong   \n",
       "4  223.60771  paid                 Bakersfield, CA    PUT  NextSong   \n",
       "\n",
       "   registration  sessionId                           song  status  \\\n",
       "0  1.538173e+12         29                      Rockpools     200   \n",
       "1  1.538332e+12          8                         Canada     200   \n",
       "2  1.538173e+12         29              Time For Miracles     200   \n",
       "3  1.538332e+12          8    Knocking On Forbidden Doors     200   \n",
       "4  1.538173e+12         29  Harder Better Faster Stronger     200   \n",
       "\n",
       "              ts                                          userAgent userId  \n",
       "0  1538352117000  Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...     30  \n",
       "1  1538352180000  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...      9  \n",
       "2  1538352394000  Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...     30  \n",
       "3  1538352416000  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...      9  \n",
       "4  1538352676000  Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...     30  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas = df.toPandas()\n",
    "df_pandas.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 286500 entries, 0 to 286499\n",
      "Data columns (total 18 columns):\n",
      "artist           228108 non-null object\n",
      "auth             286500 non-null object\n",
      "firstName        278154 non-null object\n",
      "gender           278154 non-null object\n",
      "itemInSession    286500 non-null int64\n",
      "lastName         278154 non-null object\n",
      "length           228108 non-null float64\n",
      "level            286500 non-null object\n",
      "location         278154 non-null object\n",
      "method           286500 non-null object\n",
      "page             286500 non-null object\n",
      "registration     278154 non-null float64\n",
      "sessionId        286500 non-null int64\n",
      "song             228108 non-null object\n",
      "status           286500 non-null int64\n",
      "ts               286500 non-null int64\n",
      "userAgent        278154 non-null object\n",
      "userId           286500 non-null object\n",
      "dtypes: float64(2), int64(4), object(12)\n",
      "memory usage: 39.3+ MB\n"
     ]
    }
   ],
   "source": [
    "# To know nulls\n",
    "df_pandas.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Dealing With Missing Values"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Before proceeding the data cleaning, lets see the NaN values in the dataset columns.."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "artist           58392\n",
       "auth                 0\n",
       "firstName         8346\n",
       "gender            8346\n",
       "itemInSession        0\n",
       "lastName          8346\n",
       "length           58392\n",
       "level                0\n",
       "location          8346\n",
       "method               0\n",
       "page                 0\n",
       "registration      8346\n",
       "sessionId            0\n",
       "song             58392\n",
       "status               0\n",
       "ts                   0\n",
       "userAgent         8346\n",
       "userId               0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.dropna(how = 'any', subset = ['userId', 'sessionId'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+\n",
      "|userId|\n",
      "+------+\n",
      "|      |\n",
      "|    10|\n",
      "|   100|\n",
      "|100001|\n",
      "|100002|\n",
      "+------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df.select('userId').dropDuplicates().sort('userId').show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pandas = df.toPandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "artist           58392\n",
       "auth                 0\n",
       "firstName         8346\n",
       "gender            8346\n",
       "itemInSession        0\n",
       "lastName          8346\n",
       "length           58392\n",
       "level                0\n",
       "location          8346\n",
       "method               0\n",
       "page                 0\n",
       "registration      8346\n",
       "sessionId            0\n",
       "song             58392\n",
       "status               0\n",
       "ts                   0\n",
       "userAgent         8346\n",
       "userId               0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas.isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It is seen that *FirstName, Gender, LastName, Location, Registration, and UserAgent* have the same number of missing values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['Logged In', 'Logged Out', 'Cancelled', 'Guest'], dtype=object)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas['auth'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['M', 'F', None], dtype=object)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas['gender'].unique()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It is seen that we have None values in gender section."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['paid', 'free'], dtype=object)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas['level'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['PUT', 'GET'], dtype=object)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas['method'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['NextSong', 'Add to Playlist', 'Roll Advert', 'Thumbs Up',\n",
       "       'Downgrade', 'Thumbs Down', 'Home', 'Logout', 'Help', 'Login',\n",
       "       'Upgrade', 'Add Friend', 'About', 'Settings', 'Submit Upgrade',\n",
       "       'Submit Downgrade', 'Error', 'Save Settings', 'Cancel',\n",
       "       'Cancellation Confirmation', 'Register', 'Submit Registration'], dtype=object)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas['page'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([200, 307, 404])"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#df_pandas['registration'].unique()\n",
    "df_pandas['status'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['30', '9', '74', '54', '', '4', '101', '78', '88', '95', '25',\n",
       "       '142', '13', '45', '39', '87', '28', '131', '126', '146', '59',\n",
       "       '15', '36', '120', '92', '33', '51', '27', '143', '140', '6', '124',\n",
       "       '141', '93', '81', '40', '94', '113', '98', '137', '65', '53', '42',\n",
       "       '115', '38', '58', '77', '52', '29', '70', '61', '46', '35', '132',\n",
       "       '127', '136', '80', '110', '129', '121', '82', '100', '117', '18',\n",
       "       '85', '118', '105', '96', '20', '55', '3', '12', '41', '73', '17',\n",
       "       '86', '7', '67', '32', '5', '106', '114', '69', '99', '16', '49',\n",
       "       '50', '24', '103', '102', '122', '75', '135', '128', '111', '112',\n",
       "       '145', '66', '83', '44', '11', '91', '37', '90', '104', '138',\n",
       "       '148', '107', '97', '109', '134', '10', '2', '144', '76', '119',\n",
       "       '89', '14', '62', '19', '150', '68', '147', '8', '56', '34', '60',\n",
       "       '125', '47', '133', '26', '43', '57', '149', '64', '21', '71',\n",
       "       '108', '79', '123', '116', '152', '63', '139', '23', '154', '155',\n",
       "       '72', '151', '153', '22', '84', '156', '100009', '100022', '100001',\n",
       "       '100025', '100005', '100016', '100011', '100015', '100007',\n",
       "       '100013', '100004', '100012', '100023', '100024', '100018',\n",
       "       '100006', '100019', '100010', '100002', '100003', '100008',\n",
       "       '100014', '100021', '100017', '200014', '200019', '200002',\n",
       "       '200023', '200001', '200020', '200022', '200006', '200011',\n",
       "       '200010', '200015', '200004', '200021', '200017', '200008',\n",
       "       '200024', '200003', '200018', '200009', '200025', '200016',\n",
       "       '200007', '200005', '200012', '300001', '300017', '300015',\n",
       "       '300011', '300008', '300023', '300021', '300009', '300025',\n",
       "       '300002', '300012', '300007', '300005', '300010', '300014',\n",
       "       '300016', '300019', '300020', '300022', '300013', '300018',\n",
       "       '300006', '300003', '300004', '300024'], dtype=object)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas['userId'].unique()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It is seen that we have **' '** value as an entry in the dataset. So we can clear them first and then check the None values in the gender section."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8346"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas[df_pandas['userId'] == '']['userId'].count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lets check the count in the spark dataframe.\n",
    "\n",
    "[1] https://docs.databricks.com/spark/latest/dataframes-datasets/introduction-to-dataframes-python.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8346"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.filter(df.userId == '').count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next step will be removing the **' '** values in UserId. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.filter(df.userId != '')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.filter(df.userId == '').count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "artist           50046\n",
       "auth                 0\n",
       "firstName            0\n",
       "gender               0\n",
       "itemInSession        0\n",
       "lastName             0\n",
       "length           50046\n",
       "level                0\n",
       "location             0\n",
       "method               0\n",
       "page                 0\n",
       "registration         0\n",
       "sessionId            0\n",
       "song             50046\n",
       "status               0\n",
       "ts                   0\n",
       "userAgent            0\n",
       "userId               0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas = df.toPandas()\n",
    "df_pandas.isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's observe the missing values after the conversion."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "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>artist</th>\n",
       "      <th>auth</th>\n",
       "      <th>firstName</th>\n",
       "      <th>gender</th>\n",
       "      <th>itemInSession</th>\n",
       "      <th>lastName</th>\n",
       "      <th>length</th>\n",
       "      <th>level</th>\n",
       "      <th>location</th>\n",
       "      <th>method</th>\n",
       "      <th>page</th>\n",
       "      <th>registration</th>\n",
       "      <th>sessionId</th>\n",
       "      <th>song</th>\n",
       "      <th>status</th>\n",
       "      <th>ts</th>\n",
       "      <th>userAgent</th>\n",
       "      <th>userId</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Colin</td>\n",
       "      <td>M</td>\n",
       "      <td>54</td>\n",
       "      <td>Freeman</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>Bakersfield, CA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Add to Playlist</td>\n",
       "      <td>1538173362000</td>\n",
       "      <td>29</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352905000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>84</td>\n",
       "      <td>Long</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>GET</td>\n",
       "      <td>Roll Advert</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538353150000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>86</td>\n",
       "      <td>Long</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1538353376000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Alexi</td>\n",
       "      <td>F</td>\n",
       "      <td>4</td>\n",
       "      <td>Warren</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>Spokane-Spokane Valley, WA</td>\n",
       "      <td>GET</td>\n",
       "      <td>Downgrade</td>\n",
       "      <td>1532482662000</td>\n",
       "      <td>53</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538354749000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:32.0) G...</td>\n",
       "      <td>54</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Alexi</td>\n",
       "      <td>F</td>\n",
       "      <td>7</td>\n",
       "      <td>Warren</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>Spokane-Spokane Valley, WA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1532482662000</td>\n",
       "      <td>53</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1538355255000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:32.0) G...</td>\n",
       "      <td>54</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>95</td>\n",
       "      <td>Long</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Down</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1538355306000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>97</td>\n",
       "      <td>Long</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538355504000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>99</td>\n",
       "      <td>Long</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Logout</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1538355687000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>56</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Ashlynn</td>\n",
       "      <td>F</td>\n",
       "      <td>9</td>\n",
       "      <td>Williams</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Tallahassee, FL</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1537365219000</td>\n",
       "      <td>217</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1538355711000</td>\n",
       "      <td>\"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...</td>\n",
       "      <td>74</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>57</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>104</td>\n",
       "      <td>Long</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538355855000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>67</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Colin</td>\n",
       "      <td>M</td>\n",
       "      <td>71</td>\n",
       "      <td>Freeman</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>Bakersfield, CA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Down</td>\n",
       "      <td>1538173362000</td>\n",
       "      <td>29</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1538356524000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>71</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Warren</td>\n",
       "      <td>M</td>\n",
       "      <td>1</td>\n",
       "      <td>Miller</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Baltimore-Columbia-Towson, MD</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Logout</td>\n",
       "      <td>1538169823000</td>\n",
       "      <td>240</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1538356651000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.0; rv:31.0) Gecko/20...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Warren</td>\n",
       "      <td>M</td>\n",
       "      <td>4</td>\n",
       "      <td>Miller</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Baltimore-Columbia-Towson, MD</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1538169823000</td>\n",
       "      <td>240</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538356686000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.0; rv:31.0) Gecko/20...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Ashlynn</td>\n",
       "      <td>F</td>\n",
       "      <td>15</td>\n",
       "      <td>Williams</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Tallahassee, FL</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Add to Playlist</td>\n",
       "      <td>1537365219000</td>\n",
       "      <td>217</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538356816000</td>\n",
       "      <td>\"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...</td>\n",
       "      <td>74</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Warren</td>\n",
       "      <td>M</td>\n",
       "      <td>6</td>\n",
       "      <td>Miller</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Baltimore-Columbia-Towson, MD</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1538169823000</td>\n",
       "      <td>240</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538356954000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.0; rv:31.0) Gecko/20...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>82</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Ashlynn</td>\n",
       "      <td>F</td>\n",
       "      <td>18</td>\n",
       "      <td>Williams</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Tallahassee, FL</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1537365219000</td>\n",
       "      <td>217</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1538357104000</td>\n",
       "      <td>\"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...</td>\n",
       "      <td>74</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>111</td>\n",
       "      <td>Long</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1538357160000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Colin</td>\n",
       "      <td>M</td>\n",
       "      <td>77</td>\n",
       "      <td>Freeman</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>Bakersfield, CA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1538173362000</td>\n",
       "      <td>29</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1538357649000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Ashlynn</td>\n",
       "      <td>F</td>\n",
       "      <td>22</td>\n",
       "      <td>Williams</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Tallahassee, FL</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1537365219000</td>\n",
       "      <td>217</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1538357916000</td>\n",
       "      <td>\"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...</td>\n",
       "      <td>74</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Colin</td>\n",
       "      <td>M</td>\n",
       "      <td>79</td>\n",
       "      <td>Freeman</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>Bakersfield, CA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1538173362000</td>\n",
       "      <td>29</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1538357991000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>105</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Ashlynn</td>\n",
       "      <td>F</td>\n",
       "      <td>24</td>\n",
       "      <td>Williams</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Tallahassee, FL</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Down</td>\n",
       "      <td>1537365219000</td>\n",
       "      <td>217</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1538358192000</td>\n",
       "      <td>\"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...</td>\n",
       "      <td>74</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>111</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Alex</td>\n",
       "      <td>M</td>\n",
       "      <td>1</td>\n",
       "      <td>Hogan</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Denver-Aurora-Lakewood, CO</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1535066380000</td>\n",
       "      <td>100</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538358605000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.2; WOW64; rv:31.0) G...</td>\n",
       "      <td>101</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>119</td>\n",
       "      <td>Long</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>GET</td>\n",
       "      <td>Roll Advert</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538358779000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>120</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Ainsley</td>\n",
       "      <td>F</td>\n",
       "      <td>0</td>\n",
       "      <td>Farley</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>McAllen-Edinburg-Mission, TX</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1538304455000</td>\n",
       "      <td>187</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538359181000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>78</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>123</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>122</td>\n",
       "      <td>Long</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Add to Playlist</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538359261000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>132</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Alex</td>\n",
       "      <td>M</td>\n",
       "      <td>7</td>\n",
       "      <td>Hogan</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Denver-Aurora-Lakewood, CO</td>\n",
       "      <td>GET</td>\n",
       "      <td>Help</td>\n",
       "      <td>1535066380000</td>\n",
       "      <td>100</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538359923000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.2; WOW64; rv:31.0) G...</td>\n",
       "      <td>101</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>137</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>124</td>\n",
       "      <td>Long</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>GET</td>\n",
       "      <td>Upgrade</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538360042000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>138</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>125</td>\n",
       "      <td>Long</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>GET</td>\n",
       "      <td>Upgrade</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538360044000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>139</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Zyonna</td>\n",
       "      <td>F</td>\n",
       "      <td>1</td>\n",
       "      <td>Bird</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Columbus, GA-AL</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1536663902000</td>\n",
       "      <td>241</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538360138000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...</td>\n",
       "      <td>88</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>144</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>127</td>\n",
       "      <td>Long</td>\n",
       "      <td>NaN</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>GET</td>\n",
       "      <td>Roll Advert</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1538360228000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278013</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Joshua</td>\n",
       "      <td>M</td>\n",
       "      <td>375</td>\n",
       "      <td>White</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>San Antonio-New Braunfels, TX</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1528780738000</td>\n",
       "      <td>492</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543593541000</td>\n",
       "      <td>\"Mozilla/5.0 (iPhone; CPU iPhone OS 7_1_2 like...</td>\n",
       "      <td>300015</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278021</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emilia</td>\n",
       "      <td>F</td>\n",
       "      <td>8</td>\n",
       "      <td>House</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>New York-Newark-Jersey City, NY-NJ-PA</td>\n",
       "      <td>GET</td>\n",
       "      <td>Settings</td>\n",
       "      <td>1538336771000</td>\n",
       "      <td>496</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1543594576000</td>\n",
       "      <td>Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...</td>\n",
       "      <td>300011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278024</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Joshua</td>\n",
       "      <td>M</td>\n",
       "      <td>382</td>\n",
       "      <td>White</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>San Antonio-New Braunfels, TX</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1528780738000</td>\n",
       "      <td>492</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543594794000</td>\n",
       "      <td>\"Mozilla/5.0 (iPhone; CPU iPhone OS 7_1_2 like...</td>\n",
       "      <td>300015</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278040</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Joshua</td>\n",
       "      <td>M</td>\n",
       "      <td>389</td>\n",
       "      <td>White</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>San Antonio-New Braunfels, TX</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Add Friend</td>\n",
       "      <td>1528780738000</td>\n",
       "      <td>492</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543596785000</td>\n",
       "      <td>\"Mozilla/5.0 (iPhone; CPU iPhone OS 7_1_2 like...</td>\n",
       "      <td>300015</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278041</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Joshua</td>\n",
       "      <td>M</td>\n",
       "      <td>390</td>\n",
       "      <td>White</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>San Antonio-New Braunfels, TX</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Add Friend</td>\n",
       "      <td>1528780738000</td>\n",
       "      <td>492</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543596786000</td>\n",
       "      <td>\"Mozilla/5.0 (iPhone; CPU iPhone OS 7_1_2 like...</td>\n",
       "      <td>300015</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278054</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Lauren</td>\n",
       "      <td>F</td>\n",
       "      <td>12</td>\n",
       "      <td>Boone</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>St. Louis, MO-IL</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1534859694000</td>\n",
       "      <td>485</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543600395000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>300009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278059</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Lauren</td>\n",
       "      <td>F</td>\n",
       "      <td>17</td>\n",
       "      <td>Boone</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>St. Louis, MO-IL</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1534859694000</td>\n",
       "      <td>485</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543601677000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>300009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278060</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Lauren</td>\n",
       "      <td>F</td>\n",
       "      <td>18</td>\n",
       "      <td>Boone</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>St. Louis, MO-IL</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Add to Playlist</td>\n",
       "      <td>1534859694000</td>\n",
       "      <td>485</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1543601693000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>300009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278064</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Lauren</td>\n",
       "      <td>F</td>\n",
       "      <td>22</td>\n",
       "      <td>Boone</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>St. Louis, MO-IL</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1534859694000</td>\n",
       "      <td>485</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543602458000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>300009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278068</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Lauren</td>\n",
       "      <td>F</td>\n",
       "      <td>26</td>\n",
       "      <td>Boone</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>St. Louis, MO-IL</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1534859694000</td>\n",
       "      <td>485</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543603099000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>300009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278074</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Lauren</td>\n",
       "      <td>F</td>\n",
       "      <td>31</td>\n",
       "      <td>Boone</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>St. Louis, MO-IL</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Logout</td>\n",
       "      <td>1534859694000</td>\n",
       "      <td>485</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543604084000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>300009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278075</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Lauren</td>\n",
       "      <td>F</td>\n",
       "      <td>35</td>\n",
       "      <td>Boone</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>St. Louis, MO-IL</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1534859694000</td>\n",
       "      <td>485</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1543604239000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>300009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278076</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Lauren</td>\n",
       "      <td>F</td>\n",
       "      <td>36</td>\n",
       "      <td>Boone</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>St. Louis, MO-IL</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Add Friend</td>\n",
       "      <td>1534859694000</td>\n",
       "      <td>485</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543604240000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>300009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278080</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emma</td>\n",
       "      <td>F</td>\n",
       "      <td>3</td>\n",
       "      <td>Allen</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>Wilson, NC</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Logout</td>\n",
       "      <td>1537611935000</td>\n",
       "      <td>498</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543604493000</td>\n",
       "      <td>\"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...</td>\n",
       "      <td>300021</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278083</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emma</td>\n",
       "      <td>F</td>\n",
       "      <td>7</td>\n",
       "      <td>Allen</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>Wilson, NC</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1537611935000</td>\n",
       "      <td>498</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1543604786000</td>\n",
       "      <td>\"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...</td>\n",
       "      <td>300021</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278085</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Lauren</td>\n",
       "      <td>F</td>\n",
       "      <td>41</td>\n",
       "      <td>Boone</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>St. Louis, MO-IL</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1534859694000</td>\n",
       "      <td>485</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543605019000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>300009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278089</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Lauren</td>\n",
       "      <td>F</td>\n",
       "      <td>43</td>\n",
       "      <td>Boone</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>St. Louis, MO-IL</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1534859694000</td>\n",
       "      <td>485</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1543605457000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>300009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278103</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emma</td>\n",
       "      <td>F</td>\n",
       "      <td>23</td>\n",
       "      <td>Allen</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>Wilson, NC</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Down</td>\n",
       "      <td>1537611935000</td>\n",
       "      <td>498</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543608481000</td>\n",
       "      <td>\"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...</td>\n",
       "      <td>300021</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278107</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emma</td>\n",
       "      <td>F</td>\n",
       "      <td>27</td>\n",
       "      <td>Allen</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>Wilson, NC</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1537611935000</td>\n",
       "      <td>498</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1543609421000</td>\n",
       "      <td>\"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...</td>\n",
       "      <td>300021</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278109</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emma</td>\n",
       "      <td>F</td>\n",
       "      <td>29</td>\n",
       "      <td>Allen</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>Wilson, NC</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1537611935000</td>\n",
       "      <td>498</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1543609683000</td>\n",
       "      <td>\"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...</td>\n",
       "      <td>300021</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278113</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emilia</td>\n",
       "      <td>F</td>\n",
       "      <td>2</td>\n",
       "      <td>House</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>New York-Newark-Jersey City, NY-NJ-PA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1538336771000</td>\n",
       "      <td>500</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543614016000</td>\n",
       "      <td>Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...</td>\n",
       "      <td>300011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278116</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emilia</td>\n",
       "      <td>F</td>\n",
       "      <td>5</td>\n",
       "      <td>House</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>New York-Newark-Jersey City, NY-NJ-PA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Down</td>\n",
       "      <td>1538336771000</td>\n",
       "      <td>500</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543614528000</td>\n",
       "      <td>Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...</td>\n",
       "      <td>300011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278126</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emilia</td>\n",
       "      <td>F</td>\n",
       "      <td>15</td>\n",
       "      <td>House</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>New York-Newark-Jersey City, NY-NJ-PA</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1538336771000</td>\n",
       "      <td>500</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1543617317000</td>\n",
       "      <td>Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...</td>\n",
       "      <td>300011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278132</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emilia</td>\n",
       "      <td>F</td>\n",
       "      <td>21</td>\n",
       "      <td>House</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>New York-Newark-Jersey City, NY-NJ-PA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Thumbs Up</td>\n",
       "      <td>1538336771000</td>\n",
       "      <td>500</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543618270000</td>\n",
       "      <td>Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...</td>\n",
       "      <td>300011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278136</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emilia</td>\n",
       "      <td>F</td>\n",
       "      <td>25</td>\n",
       "      <td>House</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>New York-Newark-Jersey City, NY-NJ-PA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Add Friend</td>\n",
       "      <td>1538336771000</td>\n",
       "      <td>500</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543618841000</td>\n",
       "      <td>Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...</td>\n",
       "      <td>300011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278139</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emilia</td>\n",
       "      <td>F</td>\n",
       "      <td>28</td>\n",
       "      <td>House</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>New York-Newark-Jersey City, NY-NJ-PA</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1538336771000</td>\n",
       "      <td>500</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1543619556000</td>\n",
       "      <td>Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...</td>\n",
       "      <td>300011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278147</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emilia</td>\n",
       "      <td>F</td>\n",
       "      <td>36</td>\n",
       "      <td>House</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>New York-Newark-Jersey City, NY-NJ-PA</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1538336771000</td>\n",
       "      <td>500</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1543621485000</td>\n",
       "      <td>Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...</td>\n",
       "      <td>300011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278150</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emilia</td>\n",
       "      <td>F</td>\n",
       "      <td>39</td>\n",
       "      <td>House</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>New York-Newark-Jersey City, NY-NJ-PA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>Logout</td>\n",
       "      <td>1538336771000</td>\n",
       "      <td>500</td>\n",
       "      <td>None</td>\n",
       "      <td>307</td>\n",
       "      <td>1543622122000</td>\n",
       "      <td>Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...</td>\n",
       "      <td>300011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278151</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emilia</td>\n",
       "      <td>F</td>\n",
       "      <td>43</td>\n",
       "      <td>House</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>New York-Newark-Jersey City, NY-NJ-PA</td>\n",
       "      <td>GET</td>\n",
       "      <td>Home</td>\n",
       "      <td>1538336771000</td>\n",
       "      <td>500</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1543622248000</td>\n",
       "      <td>Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...</td>\n",
       "      <td>300011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278152</th>\n",
       "      <td>None</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Emilia</td>\n",
       "      <td>F</td>\n",
       "      <td>44</td>\n",
       "      <td>House</td>\n",
       "      <td>NaN</td>\n",
       "      <td>paid</td>\n",
       "      <td>New York-Newark-Jersey City, NY-NJ-PA</td>\n",
       "      <td>GET</td>\n",
       "      <td>About</td>\n",
       "      <td>1538336771000</td>\n",
       "      <td>500</td>\n",
       "      <td>None</td>\n",
       "      <td>200</td>\n",
       "      <td>1543622398000</td>\n",
       "      <td>Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...</td>\n",
       "      <td>300011</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>50046 rows × 18 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       artist       auth firstName gender  itemInSession  lastName  length  \\\n",
       "8        None  Logged In     Colin      M             54   Freeman     NaN   \n",
       "11       None  Logged In     Micah      M             84      Long     NaN   \n",
       "14       None  Logged In     Micah      M             86      Long     NaN   \n",
       "35       None  Logged In     Alexi      F              4    Warren     NaN   \n",
       "44       None  Logged In     Alexi      F              7    Warren     NaN   \n",
       "47       None  Logged In     Micah      M             95      Long     NaN   \n",
       "51       None  Logged In     Micah      M             97      Long     NaN   \n",
       "53       None  Logged In     Micah      M             99      Long     NaN   \n",
       "56       None  Logged In   Ashlynn      F              9  Williams     NaN   \n",
       "57       None  Logged In     Micah      M            104      Long     NaN   \n",
       "67       None  Logged In     Colin      M             71   Freeman     NaN   \n",
       "71       None  Logged In    Warren      M              1    Miller     NaN   \n",
       "72       None  Logged In    Warren      M              4    Miller     NaN   \n",
       "75       None  Logged In   Ashlynn      F             15  Williams     NaN   \n",
       "80       None  Logged In    Warren      M              6    Miller     NaN   \n",
       "82       None  Logged In   Ashlynn      F             18  Williams     NaN   \n",
       "85       None  Logged In     Micah      M            111      Long     NaN   \n",
       "95       None  Logged In     Colin      M             77   Freeman     NaN   \n",
       "99       None  Logged In   Ashlynn      F             22  Williams     NaN   \n",
       "101      None  Logged In     Colin      M             79   Freeman     NaN   \n",
       "105      None  Logged In   Ashlynn      F             24  Williams     NaN   \n",
       "111      None  Logged In      Alex      M              1     Hogan     NaN   \n",
       "115      None  Logged In     Micah      M            119      Long     NaN   \n",
       "120      None  Logged In   Ainsley      F              0    Farley     NaN   \n",
       "123      None  Logged In     Micah      M            122      Long     NaN   \n",
       "132      None  Logged In      Alex      M              7     Hogan     NaN   \n",
       "137      None  Logged In     Micah      M            124      Long     NaN   \n",
       "138      None  Logged In     Micah      M            125      Long     NaN   \n",
       "139      None  Logged In    Zyonna      F              1      Bird     NaN   \n",
       "144      None  Logged In     Micah      M            127      Long     NaN   \n",
       "...       ...        ...       ...    ...            ...       ...     ...   \n",
       "278013   None  Logged In    Joshua      M            375     White     NaN   \n",
       "278021   None  Logged In    Emilia      F              8     House     NaN   \n",
       "278024   None  Logged In    Joshua      M            382     White     NaN   \n",
       "278040   None  Logged In    Joshua      M            389     White     NaN   \n",
       "278041   None  Logged In    Joshua      M            390     White     NaN   \n",
       "278054   None  Logged In    Lauren      F             12     Boone     NaN   \n",
       "278059   None  Logged In    Lauren      F             17     Boone     NaN   \n",
       "278060   None  Logged In    Lauren      F             18     Boone     NaN   \n",
       "278064   None  Logged In    Lauren      F             22     Boone     NaN   \n",
       "278068   None  Logged In    Lauren      F             26     Boone     NaN   \n",
       "278074   None  Logged In    Lauren      F             31     Boone     NaN   \n",
       "278075   None  Logged In    Lauren      F             35     Boone     NaN   \n",
       "278076   None  Logged In    Lauren      F             36     Boone     NaN   \n",
       "278080   None  Logged In      Emma      F              3     Allen     NaN   \n",
       "278083   None  Logged In      Emma      F              7     Allen     NaN   \n",
       "278085   None  Logged In    Lauren      F             41     Boone     NaN   \n",
       "278089   None  Logged In    Lauren      F             43     Boone     NaN   \n",
       "278103   None  Logged In      Emma      F             23     Allen     NaN   \n",
       "278107   None  Logged In      Emma      F             27     Allen     NaN   \n",
       "278109   None  Logged In      Emma      F             29     Allen     NaN   \n",
       "278113   None  Logged In    Emilia      F              2     House     NaN   \n",
       "278116   None  Logged In    Emilia      F              5     House     NaN   \n",
       "278126   None  Logged In    Emilia      F             15     House     NaN   \n",
       "278132   None  Logged In    Emilia      F             21     House     NaN   \n",
       "278136   None  Logged In    Emilia      F             25     House     NaN   \n",
       "278139   None  Logged In    Emilia      F             28     House     NaN   \n",
       "278147   None  Logged In    Emilia      F             36     House     NaN   \n",
       "278150   None  Logged In    Emilia      F             39     House     NaN   \n",
       "278151   None  Logged In    Emilia      F             43     House     NaN   \n",
       "278152   None  Logged In    Emilia      F             44     House     NaN   \n",
       "\n",
       "       level                               location method             page  \\\n",
       "8       paid                        Bakersfield, CA    PUT  Add to Playlist   \n",
       "11      free         Boston-Cambridge-Newton, MA-NH    GET      Roll Advert   \n",
       "14      free         Boston-Cambridge-Newton, MA-NH    PUT        Thumbs Up   \n",
       "35      paid             Spokane-Spokane Valley, WA    GET        Downgrade   \n",
       "44      paid             Spokane-Spokane Valley, WA    PUT        Thumbs Up   \n",
       "47      free         Boston-Cambridge-Newton, MA-NH    PUT      Thumbs Down   \n",
       "51      free         Boston-Cambridge-Newton, MA-NH    GET             Home   \n",
       "53      free         Boston-Cambridge-Newton, MA-NH    PUT           Logout   \n",
       "56      free                        Tallahassee, FL    PUT        Thumbs Up   \n",
       "57      free         Boston-Cambridge-Newton, MA-NH    GET             Home   \n",
       "67      paid                        Bakersfield, CA    PUT      Thumbs Down   \n",
       "71      free          Baltimore-Columbia-Towson, MD    PUT           Logout   \n",
       "72      free          Baltimore-Columbia-Towson, MD    GET             Home   \n",
       "75      free                        Tallahassee, FL    PUT  Add to Playlist   \n",
       "80      free          Baltimore-Columbia-Towson, MD    GET             Home   \n",
       "82      free                        Tallahassee, FL    PUT        Thumbs Up   \n",
       "85      free         Boston-Cambridge-Newton, MA-NH    PUT        Thumbs Up   \n",
       "95      paid                        Bakersfield, CA    PUT        Thumbs Up   \n",
       "99      free                        Tallahassee, FL    PUT        Thumbs Up   \n",
       "101     paid                        Bakersfield, CA    PUT        Thumbs Up   \n",
       "105     free                        Tallahassee, FL    PUT      Thumbs Down   \n",
       "111     free             Denver-Aurora-Lakewood, CO    GET             Home   \n",
       "115     free         Boston-Cambridge-Newton, MA-NH    GET      Roll Advert   \n",
       "120     free           McAllen-Edinburg-Mission, TX    GET             Home   \n",
       "123     free         Boston-Cambridge-Newton, MA-NH    PUT  Add to Playlist   \n",
       "132     free             Denver-Aurora-Lakewood, CO    GET             Help   \n",
       "137     free         Boston-Cambridge-Newton, MA-NH    GET          Upgrade   \n",
       "138     free         Boston-Cambridge-Newton, MA-NH    GET          Upgrade   \n",
       "139     free                        Columbus, GA-AL    GET             Home   \n",
       "144     free         Boston-Cambridge-Newton, MA-NH    GET      Roll Advert   \n",
       "...      ...                                    ...    ...              ...   \n",
       "278013  paid          San Antonio-New Braunfels, TX    PUT        Thumbs Up   \n",
       "278021  paid  New York-Newark-Jersey City, NY-NJ-PA    GET         Settings   \n",
       "278024  paid          San Antonio-New Braunfels, TX    PUT        Thumbs Up   \n",
       "278040  paid          San Antonio-New Braunfels, TX    PUT       Add Friend   \n",
       "278041  paid          San Antonio-New Braunfels, TX    PUT       Add Friend   \n",
       "278054  paid                       St. Louis, MO-IL    PUT        Thumbs Up   \n",
       "278059  paid                       St. Louis, MO-IL    PUT        Thumbs Up   \n",
       "278060  paid                       St. Louis, MO-IL    PUT  Add to Playlist   \n",
       "278064  paid                       St. Louis, MO-IL    PUT        Thumbs Up   \n",
       "278068  paid                       St. Louis, MO-IL    PUT        Thumbs Up   \n",
       "278074  paid                       St. Louis, MO-IL    PUT           Logout   \n",
       "278075  paid                       St. Louis, MO-IL    GET             Home   \n",
       "278076  paid                       St. Louis, MO-IL    PUT       Add Friend   \n",
       "278080  paid                             Wilson, NC    PUT           Logout   \n",
       "278083  paid                             Wilson, NC    GET             Home   \n",
       "278085  paid                       St. Louis, MO-IL    PUT        Thumbs Up   \n",
       "278089  paid                       St. Louis, MO-IL    GET             Home   \n",
       "278103  paid                             Wilson, NC    PUT      Thumbs Down   \n",
       "278107  paid                             Wilson, NC    GET             Home   \n",
       "278109  paid                             Wilson, NC    GET             Home   \n",
       "278113  paid  New York-Newark-Jersey City, NY-NJ-PA    PUT        Thumbs Up   \n",
       "278116  paid  New York-Newark-Jersey City, NY-NJ-PA    PUT      Thumbs Down   \n",
       "278126  paid  New York-Newark-Jersey City, NY-NJ-PA    GET             Home   \n",
       "278132  paid  New York-Newark-Jersey City, NY-NJ-PA    PUT        Thumbs Up   \n",
       "278136  paid  New York-Newark-Jersey City, NY-NJ-PA    PUT       Add Friend   \n",
       "278139  paid  New York-Newark-Jersey City, NY-NJ-PA    GET             Home   \n",
       "278147  paid  New York-Newark-Jersey City, NY-NJ-PA    GET             Home   \n",
       "278150  paid  New York-Newark-Jersey City, NY-NJ-PA    PUT           Logout   \n",
       "278151  paid  New York-Newark-Jersey City, NY-NJ-PA    GET             Home   \n",
       "278152  paid  New York-Newark-Jersey City, NY-NJ-PA    GET            About   \n",
       "\n",
       "         registration  sessionId  song  status             ts  \\\n",
       "8       1538173362000         29  None     200  1538352905000   \n",
       "11      1538331630000          8  None     200  1538353150000   \n",
       "14      1538331630000          8  None     307  1538353376000   \n",
       "35      1532482662000         53  None     200  1538354749000   \n",
       "44      1532482662000         53  None     307  1538355255000   \n",
       "47      1538331630000          8  None     307  1538355306000   \n",
       "51      1538331630000          8  None     200  1538355504000   \n",
       "53      1538331630000          8  None     307  1538355687000   \n",
       "56      1537365219000        217  None     307  1538355711000   \n",
       "57      1538331630000          8  None     200  1538355855000   \n",
       "67      1538173362000         29  None     307  1538356524000   \n",
       "71      1538169823000        240  None     307  1538356651000   \n",
       "72      1538169823000        240  None     200  1538356686000   \n",
       "75      1537365219000        217  None     200  1538356816000   \n",
       "80      1538169823000        240  None     200  1538356954000   \n",
       "82      1537365219000        217  None     307  1538357104000   \n",
       "85      1538331630000          8  None     307  1538357160000   \n",
       "95      1538173362000         29  None     307  1538357649000   \n",
       "99      1537365219000        217  None     307  1538357916000   \n",
       "101     1538173362000         29  None     307  1538357991000   \n",
       "105     1537365219000        217  None     307  1538358192000   \n",
       "111     1535066380000        100  None     200  1538358605000   \n",
       "115     1538331630000          8  None     200  1538358779000   \n",
       "120     1538304455000        187  None     200  1538359181000   \n",
       "123     1538331630000          8  None     200  1538359261000   \n",
       "132     1535066380000        100  None     200  1538359923000   \n",
       "137     1538331630000          8  None     200  1538360042000   \n",
       "138     1538331630000          8  None     200  1538360044000   \n",
       "139     1536663902000        241  None     200  1538360138000   \n",
       "144     1538331630000          8  None     200  1538360228000   \n",
       "...               ...        ...   ...     ...            ...   \n",
       "278013  1528780738000        492  None     307  1543593541000   \n",
       "278021  1538336771000        496  None     200  1543594576000   \n",
       "278024  1528780738000        492  None     307  1543594794000   \n",
       "278040  1528780738000        492  None     307  1543596785000   \n",
       "278041  1528780738000        492  None     307  1543596786000   \n",
       "278054  1534859694000        485  None     307  1543600395000   \n",
       "278059  1534859694000        485  None     307  1543601677000   \n",
       "278060  1534859694000        485  None     200  1543601693000   \n",
       "278064  1534859694000        485  None     307  1543602458000   \n",
       "278068  1534859694000        485  None     307  1543603099000   \n",
       "278074  1534859694000        485  None     307  1543604084000   \n",
       "278075  1534859694000        485  None     200  1543604239000   \n",
       "278076  1534859694000        485  None     307  1543604240000   \n",
       "278080  1537611935000        498  None     307  1543604493000   \n",
       "278083  1537611935000        498  None     200  1543604786000   \n",
       "278085  1534859694000        485  None     307  1543605019000   \n",
       "278089  1534859694000        485  None     200  1543605457000   \n",
       "278103  1537611935000        498  None     307  1543608481000   \n",
       "278107  1537611935000        498  None     200  1543609421000   \n",
       "278109  1537611935000        498  None     200  1543609683000   \n",
       "278113  1538336771000        500  None     307  1543614016000   \n",
       "278116  1538336771000        500  None     307  1543614528000   \n",
       "278126  1538336771000        500  None     200  1543617317000   \n",
       "278132  1538336771000        500  None     307  1543618270000   \n",
       "278136  1538336771000        500  None     307  1543618841000   \n",
       "278139  1538336771000        500  None     200  1543619556000   \n",
       "278147  1538336771000        500  None     200  1543621485000   \n",
       "278150  1538336771000        500  None     307  1543622122000   \n",
       "278151  1538336771000        500  None     200  1543622248000   \n",
       "278152  1538336771000        500  None     200  1543622398000   \n",
       "\n",
       "                                                userAgent  userId  \n",
       "8       Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...      30  \n",
       "11      \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...       9  \n",
       "14      \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...       9  \n",
       "35      Mozilla/5.0 (Windows NT 6.1; WOW64; rv:32.0) G...      54  \n",
       "44      Mozilla/5.0 (Windows NT 6.1; WOW64; rv:32.0) G...      54  \n",
       "47      \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...       9  \n",
       "51      \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...       9  \n",
       "53      \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...       9  \n",
       "56      \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...      74  \n",
       "57      \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...       9  \n",
       "67      Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...      30  \n",
       "71      Mozilla/5.0 (Windows NT 6.0; rv:31.0) Gecko/20...       4  \n",
       "72      Mozilla/5.0 (Windows NT 6.0; rv:31.0) Gecko/20...       4  \n",
       "75      \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...      74  \n",
       "80      Mozilla/5.0 (Windows NT 6.0; rv:31.0) Gecko/20...       4  \n",
       "82      \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...      74  \n",
       "85      \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...       9  \n",
       "95      Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...      30  \n",
       "99      \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...      74  \n",
       "101     Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...      30  \n",
       "105     \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...      74  \n",
       "111     Mozilla/5.0 (Windows NT 6.2; WOW64; rv:31.0) G...     101  \n",
       "115     \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...       9  \n",
       "120     \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...      78  \n",
       "123     \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...       9  \n",
       "132     Mozilla/5.0 (Windows NT 6.2; WOW64; rv:31.0) G...     101  \n",
       "137     \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...       9  \n",
       "138     \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...       9  \n",
       "139     Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...      88  \n",
       "144     \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...       9  \n",
       "...                                                   ...     ...  \n",
       "278013  \"Mozilla/5.0 (iPhone; CPU iPhone OS 7_1_2 like...  300015  \n",
       "278021  Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...  300011  \n",
       "278024  \"Mozilla/5.0 (iPhone; CPU iPhone OS 7_1_2 like...  300015  \n",
       "278040  \"Mozilla/5.0 (iPhone; CPU iPhone OS 7_1_2 like...  300015  \n",
       "278041  \"Mozilla/5.0 (iPhone; CPU iPhone OS 7_1_2 like...  300015  \n",
       "278054  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...  300009  \n",
       "278059  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...  300009  \n",
       "278060  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...  300009  \n",
       "278064  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...  300009  \n",
       "278068  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...  300009  \n",
       "278074  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...  300009  \n",
       "278075  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...  300009  \n",
       "278076  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...  300009  \n",
       "278080  \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...  300021  \n",
       "278083  \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...  300021  \n",
       "278085  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...  300009  \n",
       "278089  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...  300009  \n",
       "278103  \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...  300021  \n",
       "278107  \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...  300021  \n",
       "278109  \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...  300021  \n",
       "278113  Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...  300011  \n",
       "278116  Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...  300011  \n",
       "278126  Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...  300011  \n",
       "278132  Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...  300011  \n",
       "278136  Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...  300011  \n",
       "278139  Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...  300011  \n",
       "278147  Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...  300011  \n",
       "278150  Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...  300011  \n",
       "278151  Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...  300011  \n",
       "278152  Mozilla/5.0 (compatible; MSIE 9.0; Windows NT ...  300011  \n",
       "\n",
       "[50046 rows x 18 columns]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas[df_pandas.isnull().any(axis=1)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Since we will make analysis based on Chunk, it is not needed to drop the None values in artist, length, and song column. Because, they indicate that user is logged in but not listening any song yet therefore no length value."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Unique values in auth  column : \n",
      "['Logged In' 'Cancelled']\n",
      "Unique values in gender  column : \n",
      "['M' 'F']\n",
      "Unique values in level  column : \n",
      "['paid' 'free']\n",
      "Unique values in method  column : \n",
      "['PUT' 'GET']\n",
      "Unique values in page  column : \n",
      "['NextSong' 'Add to Playlist' 'Roll Advert' 'Thumbs Up' 'Downgrade'\n",
      " 'Thumbs Down' 'Home' 'Logout' 'Help' 'Upgrade' 'Add Friend' 'Settings'\n",
      " 'Submit Upgrade' 'About' 'Submit Downgrade' 'Error' 'Save Settings'\n",
      " 'Cancel' 'Cancellation Confirmation']\n",
      "Unique values in status  column : \n",
      "[200 307 404]\n"
     ]
    }
   ],
   "source": [
    "print('Unique values in auth  column : \\n{}'.format(df_pandas['auth'].unique()))\n",
    "print('Unique values in gender  column : \\n{}'.format(df_pandas['gender'].unique()))\n",
    "print('Unique values in level  column : \\n{}'.format(df_pandas['level'].unique()))\n",
    "print('Unique values in method  column : \\n{}'.format(df_pandas['method'].unique()))\n",
    "print('Unique values in page  column : \\n{}'.format(df_pandas['page'].unique()))\n",
    "print('Unique values in status  column : \\n{}'.format(df_pandas['status'].unique()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It is seen that after dropping the ' ' values in UserId column, we also clean the FirstName, Gender, LastName, Location, Registration, and UserAgent columns which have the same missing number count as UserId ' ' count."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "When we compare the unique values in dataset before and after the cleaning of UserId column, it is seen that we have lost two information in the auth column, Logged out and Guest. Additionaly, in the page column, we do not have Login, Registration, and Submit Registration informations anymore. \n",
    "\n",
    "Since we have dropped the *empty* user ıds, it might be logical in the sense of users without userid can be guests or logged out users. It make sense for page column also because the lost informations are related to guest users and logged out users.\n",
    "\n",
    "It is also logical that we have lost None values in gender column, because guest users or logged out users do not have gender information. \n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Converting The Timestamp"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "[1] https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.Timestamp.html\n",
    "\n",
    "[2] https://stackoverflow.com/questions/13890935/does-pythons-time-time-return-the-local-or-utc-timestamp\n",
    "\n",
    "[3] https://stackoverflow.com/questions/2265357/parse-date-string-and-change-format"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    1538352117000\n",
       "1    1538352180000\n",
       "2    1538352394000\n",
       "3    1538352416000\n",
       "4    1538352676000\n",
       "Name: ts, dtype: int64"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas['ts'].head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "get_time = udf(lambda x: datetime.datetime.fromtimestamp(x / 1000.0).strftime(\"%Y-%m-%d %H:%M:%S\"))\n",
    "df = df.withColumn(\"time\", get_time(df.ts))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In order to use Python function in Spark, I have used udf.\n",
    "\n",
    "[1] https://changhsinlee.com/pyspark-udf/\n",
    "\n",
    "[2] https://docs.databricks.com/spark/latest/spark-sql/udf-python.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pandas = df.toPandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "scrolled": true
   },
   "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>artist</th>\n",
       "      <th>auth</th>\n",
       "      <th>firstName</th>\n",
       "      <th>gender</th>\n",
       "      <th>itemInSession</th>\n",
       "      <th>lastName</th>\n",
       "      <th>length</th>\n",
       "      <th>level</th>\n",
       "      <th>location</th>\n",
       "      <th>method</th>\n",
       "      <th>page</th>\n",
       "      <th>registration</th>\n",
       "      <th>sessionId</th>\n",
       "      <th>song</th>\n",
       "      <th>status</th>\n",
       "      <th>ts</th>\n",
       "      <th>userAgent</th>\n",
       "      <th>userId</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Martha Tilston</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Colin</td>\n",
       "      <td>M</td>\n",
       "      <td>50</td>\n",
       "      <td>Freeman</td>\n",
       "      <td>277.89016</td>\n",
       "      <td>paid</td>\n",
       "      <td>Bakersfield, CA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1538173362000</td>\n",
       "      <td>29</td>\n",
       "      <td>Rockpools</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352117000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...</td>\n",
       "      <td>30</td>\n",
       "      <td>2018-10-01 00:01:57</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Five Iron Frenzy</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>79</td>\n",
       "      <td>Long</td>\n",
       "      <td>236.09424</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>Canada</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352180000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "      <td>2018-10-01 00:03:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Adam Lambert</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Colin</td>\n",
       "      <td>M</td>\n",
       "      <td>51</td>\n",
       "      <td>Freeman</td>\n",
       "      <td>282.82730</td>\n",
       "      <td>paid</td>\n",
       "      <td>Bakersfield, CA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1538173362000</td>\n",
       "      <td>29</td>\n",
       "      <td>Time For Miracles</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352394000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...</td>\n",
       "      <td>30</td>\n",
       "      <td>2018-10-01 00:06:34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Enigma</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>80</td>\n",
       "      <td>Long</td>\n",
       "      <td>262.71302</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>Knocking On Forbidden Doors</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352416000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "      <td>2018-10-01 00:06:56</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Daft Punk</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Colin</td>\n",
       "      <td>M</td>\n",
       "      <td>52</td>\n",
       "      <td>Freeman</td>\n",
       "      <td>223.60771</td>\n",
       "      <td>paid</td>\n",
       "      <td>Bakersfield, CA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1538173362000</td>\n",
       "      <td>29</td>\n",
       "      <td>Harder Better Faster Stronger</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352676000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...</td>\n",
       "      <td>30</td>\n",
       "      <td>2018-10-01 00:11:16</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             artist       auth firstName gender  itemInSession lastName  \\\n",
       "0    Martha Tilston  Logged In     Colin      M             50  Freeman   \n",
       "1  Five Iron Frenzy  Logged In     Micah      M             79     Long   \n",
       "2      Adam Lambert  Logged In     Colin      M             51  Freeman   \n",
       "3            Enigma  Logged In     Micah      M             80     Long   \n",
       "4         Daft Punk  Logged In     Colin      M             52  Freeman   \n",
       "\n",
       "      length level                        location method      page  \\\n",
       "0  277.89016  paid                 Bakersfield, CA    PUT  NextSong   \n",
       "1  236.09424  free  Boston-Cambridge-Newton, MA-NH    PUT  NextSong   \n",
       "2  282.82730  paid                 Bakersfield, CA    PUT  NextSong   \n",
       "3  262.71302  free  Boston-Cambridge-Newton, MA-NH    PUT  NextSong   \n",
       "4  223.60771  paid                 Bakersfield, CA    PUT  NextSong   \n",
       "\n",
       "    registration  sessionId                           song  status  \\\n",
       "0  1538173362000         29                      Rockpools     200   \n",
       "1  1538331630000          8                         Canada     200   \n",
       "2  1538173362000         29              Time For Miracles     200   \n",
       "3  1538331630000          8    Knocking On Forbidden Doors     200   \n",
       "4  1538173362000         29  Harder Better Faster Stronger     200   \n",
       "\n",
       "              ts                                          userAgent userId  \\\n",
       "0  1538352117000  Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...     30   \n",
       "1  1538352180000  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...      9   \n",
       "2  1538352394000  Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...     30   \n",
       "3  1538352416000  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...      9   \n",
       "4  1538352676000  Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...     30   \n",
       "\n",
       "                  time  \n",
       "0  2018-10-01 00:01:57  \n",
       "1  2018-10-01 00:03:00  \n",
       "2  2018-10-01 00:06:34  \n",
       "3  2018-10-01 00:06:56  \n",
       "4  2018-10-01 00:11:16  "
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Since the registration column is also in timestamp format, I will perform the same conversion of ts column for registration column."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.withColumn('registration_time', get_time(df.registration))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pandas = df.toPandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "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>artist</th>\n",
       "      <th>auth</th>\n",
       "      <th>firstName</th>\n",
       "      <th>gender</th>\n",
       "      <th>itemInSession</th>\n",
       "      <th>lastName</th>\n",
       "      <th>length</th>\n",
       "      <th>level</th>\n",
       "      <th>location</th>\n",
       "      <th>method</th>\n",
       "      <th>page</th>\n",
       "      <th>registration</th>\n",
       "      <th>sessionId</th>\n",
       "      <th>song</th>\n",
       "      <th>status</th>\n",
       "      <th>ts</th>\n",
       "      <th>userAgent</th>\n",
       "      <th>userId</th>\n",
       "      <th>time</th>\n",
       "      <th>registration_time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Martha Tilston</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Colin</td>\n",
       "      <td>M</td>\n",
       "      <td>50</td>\n",
       "      <td>Freeman</td>\n",
       "      <td>277.89016</td>\n",
       "      <td>paid</td>\n",
       "      <td>Bakersfield, CA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1538173362000</td>\n",
       "      <td>29</td>\n",
       "      <td>Rockpools</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352117000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...</td>\n",
       "      <td>30</td>\n",
       "      <td>2018-10-01 00:01:57</td>\n",
       "      <td>2018-09-28 22:22:42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Five Iron Frenzy</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>79</td>\n",
       "      <td>Long</td>\n",
       "      <td>236.09424</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>Canada</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352180000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "      <td>2018-10-01 00:03:00</td>\n",
       "      <td>2018-09-30 18:20:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Adam Lambert</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Colin</td>\n",
       "      <td>M</td>\n",
       "      <td>51</td>\n",
       "      <td>Freeman</td>\n",
       "      <td>282.82730</td>\n",
       "      <td>paid</td>\n",
       "      <td>Bakersfield, CA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1538173362000</td>\n",
       "      <td>29</td>\n",
       "      <td>Time For Miracles</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352394000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...</td>\n",
       "      <td>30</td>\n",
       "      <td>2018-10-01 00:06:34</td>\n",
       "      <td>2018-09-28 22:22:42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Enigma</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Micah</td>\n",
       "      <td>M</td>\n",
       "      <td>80</td>\n",
       "      <td>Long</td>\n",
       "      <td>262.71302</td>\n",
       "      <td>free</td>\n",
       "      <td>Boston-Cambridge-Newton, MA-NH</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1538331630000</td>\n",
       "      <td>8</td>\n",
       "      <td>Knocking On Forbidden Doors</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352416000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...</td>\n",
       "      <td>9</td>\n",
       "      <td>2018-10-01 00:06:56</td>\n",
       "      <td>2018-09-30 18:20:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Daft Punk</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Colin</td>\n",
       "      <td>M</td>\n",
       "      <td>52</td>\n",
       "      <td>Freeman</td>\n",
       "      <td>223.60771</td>\n",
       "      <td>paid</td>\n",
       "      <td>Bakersfield, CA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>NextSong</td>\n",
       "      <td>1538173362000</td>\n",
       "      <td>29</td>\n",
       "      <td>Harder Better Faster Stronger</td>\n",
       "      <td>200</td>\n",
       "      <td>1538352676000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...</td>\n",
       "      <td>30</td>\n",
       "      <td>2018-10-01 00:11:16</td>\n",
       "      <td>2018-09-28 22:22:42</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             artist       auth firstName gender  itemInSession lastName  \\\n",
       "0    Martha Tilston  Logged In     Colin      M             50  Freeman   \n",
       "1  Five Iron Frenzy  Logged In     Micah      M             79     Long   \n",
       "2      Adam Lambert  Logged In     Colin      M             51  Freeman   \n",
       "3            Enigma  Logged In     Micah      M             80     Long   \n",
       "4         Daft Punk  Logged In     Colin      M             52  Freeman   \n",
       "\n",
       "      length level                        location method      page  \\\n",
       "0  277.89016  paid                 Bakersfield, CA    PUT  NextSong   \n",
       "1  236.09424  free  Boston-Cambridge-Newton, MA-NH    PUT  NextSong   \n",
       "2  282.82730  paid                 Bakersfield, CA    PUT  NextSong   \n",
       "3  262.71302  free  Boston-Cambridge-Newton, MA-NH    PUT  NextSong   \n",
       "4  223.60771  paid                 Bakersfield, CA    PUT  NextSong   \n",
       "\n",
       "    registration  sessionId                           song  status  \\\n",
       "0  1538173362000         29                      Rockpools     200   \n",
       "1  1538331630000          8                         Canada     200   \n",
       "2  1538173362000         29              Time For Miracles     200   \n",
       "3  1538331630000          8    Knocking On Forbidden Doors     200   \n",
       "4  1538173362000         29  Harder Better Faster Stronger     200   \n",
       "\n",
       "              ts                                          userAgent userId  \\\n",
       "0  1538352117000  Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...     30   \n",
       "1  1538352180000  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...      9   \n",
       "2  1538352394000  Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...     30   \n",
       "3  1538352416000  \"Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebK...      9   \n",
       "4  1538352676000  Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) G...     30   \n",
       "\n",
       "                  time    registration_time  \n",
       "0  2018-10-01 00:01:57  2018-09-28 22:22:42  \n",
       "1  2018-10-01 00:03:00  2018-09-30 18:20:30  \n",
       "2  2018-10-01 00:06:34  2018-09-28 22:22:42  \n",
       "3  2018-10-01 00:06:56  2018-09-30 18:20:30  \n",
       "4  2018-10-01 00:11:16  2018-09-28 22:22:42  "
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Exploratory Data Analysis\n",
    "When you're working with the full dataset, perform EDA by loading a small subset of the data and doing basic manipulations within Spark. In this workspace, you are already provided a small subset of data you can explore.\n",
    "\n",
    "### Define Churn\n",
    "\n",
    "Once you've done some preliminary analysis, create a column `Churn` to use as the label for your model. I suggest using the `Cancellation Confirmation` events to define your churn, which happen for both paid and free users. As a bonus task, you can also look into the `Downgrade` events.\n",
    "\n",
    "Customer churn occurs when customers/subscribers stop doing business with a company or service. We can define a user as churned if there is an Cancellation Confirmation event. And Churn is defined both free and paid users. In order to track the downgrade event of a paid user can also be sight about customer churn.\n",
    "\n",
    "\n",
    "[1] https://blog.recurly.com/better-way-to-calculate-your-churn-rate\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First look for cancellation confirmation events to define Churn."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "# create churn label\n",
    "flag_cancellation = udf(lambda x: 1 if x == 'Cancellation Confirmation' else 0, IntegerType())\n",
    "\n",
    "# Apply it to the dataframe\n",
    "df = df.withColumn('churn', flag_cancellation('page'))\n",
    "\n",
    "# label user who churned\n",
    "windowval = Window.partitionBy('userId')\n",
    "\n",
    "# Apply it to the dataframe\n",
    "df = df.withColumn('churn', max('churn').over(windowval))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pandas = df.toPandas()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, look for downgrade events."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "# create downgrade label\n",
    "flag_downgrade = udf(lambda x: 1 if x == \"Submit Downgrade\" else 0, IntegerType())\n",
    "\n",
    "# Apply it to the dataframe\n",
    "df = df.withColumn(\"downgrade\", flag_downgrade(\"page\"))\n",
    "\n",
    "# label user who've ever downgraded\n",
    "windowval = Window.partitionBy('userId')\n",
    "\n",
    "# Apply it to the dataframe\n",
    "df = df.withColumn('downgrade', max('downgrade').over(windowval))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pandas = df.toPandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "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>artist</th>\n",
       "      <th>auth</th>\n",
       "      <th>firstName</th>\n",
       "      <th>gender</th>\n",
       "      <th>itemInSession</th>\n",
       "      <th>lastName</th>\n",
       "      <th>length</th>\n",
       "      <th>level</th>\n",
       "      <th>location</th>\n",
       "      <th>method</th>\n",
       "      <th>...</th>\n",
       "      <th>sessionId</th>\n",
       "      <th>song</th>\n",
       "      <th>status</th>\n",
       "      <th>ts</th>\n",
       "      <th>userAgent</th>\n",
       "      <th>userId</th>\n",
       "      <th>time</th>\n",
       "      <th>registration_time</th>\n",
       "      <th>churn</th>\n",
       "      <th>downgrade</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>185153</th>\n",
       "      <td>Jamie T</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Daniel</td>\n",
       "      <td>M</td>\n",
       "      <td>161</td>\n",
       "      <td>Marks</td>\n",
       "      <td>259.65669</td>\n",
       "      <td>paid</td>\n",
       "      <td>Chicago-Naperville-Elgin, IL-IN-WI</td>\n",
       "      <td>PUT</td>\n",
       "      <td>...</td>\n",
       "      <td>94</td>\n",
       "      <td>Sheila (Explicit)</td>\n",
       "      <td>200</td>\n",
       "      <td>1539972314000</td>\n",
       "      <td>\"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...</td>\n",
       "      <td>100012</td>\n",
       "      <td>2018-10-19 18:05:14</td>\n",
       "      <td>2018-09-19 18:19:14</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12368</th>\n",
       "      <td>Energy 52</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Alexi</td>\n",
       "      <td>F</td>\n",
       "      <td>30</td>\n",
       "      <td>Warren</td>\n",
       "      <td>462.13179</td>\n",
       "      <td>free</td>\n",
       "      <td>Spokane-Spokane Valley, WA</td>\n",
       "      <td>PUT</td>\n",
       "      <td>...</td>\n",
       "      <td>963</td>\n",
       "      <td>CafÃÂ© Del Mar</td>\n",
       "      <td>200</td>\n",
       "      <td>1539789477000</td>\n",
       "      <td>Mozilla/5.0 (Windows NT 6.1; WOW64; rv:32.0) G...</td>\n",
       "      <td>54</td>\n",
       "      <td>2018-10-17 15:17:57</td>\n",
       "      <td>2018-07-25 01:37:42</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>178657</th>\n",
       "      <td>P.O.D.</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Erick</td>\n",
       "      <td>M</td>\n",
       "      <td>153</td>\n",
       "      <td>Brooks</td>\n",
       "      <td>258.16771</td>\n",
       "      <td>paid</td>\n",
       "      <td>Selma, AL</td>\n",
       "      <td>PUT</td>\n",
       "      <td>...</td>\n",
       "      <td>1069</td>\n",
       "      <td>Youth Of The Nation (2006 Remastered Album Ver...</td>\n",
       "      <td>200</td>\n",
       "      <td>1540023127000</td>\n",
       "      <td>\"Mozilla/5.0 (Windows NT 6.3; WOW64) AppleWebK...</td>\n",
       "      <td>58</td>\n",
       "      <td>2018-10-20 08:12:07</td>\n",
       "      <td>2018-09-26 10:12:31</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>190882</th>\n",
       "      <td>DJ Khaled</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Kaylenn</td>\n",
       "      <td>F</td>\n",
       "      <td>82</td>\n",
       "      <td>Jenkins</td>\n",
       "      <td>227.08200</td>\n",
       "      <td>paid</td>\n",
       "      <td>Price, UT</td>\n",
       "      <td>PUT</td>\n",
       "      <td>...</td>\n",
       "      <td>649</td>\n",
       "      <td>All I Do Is Win (feat. T-Pain_ Ludacris_ Snoop...</td>\n",
       "      <td>200</td>\n",
       "      <td>1540054668000</td>\n",
       "      <td>\"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_2...</td>\n",
       "      <td>141</td>\n",
       "      <td>2018-10-20 16:57:48</td>\n",
       "      <td>2018-09-02 15:57:58</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>256047</th>\n",
       "      <td>LMFAO</td>\n",
       "      <td>Logged In</td>\n",
       "      <td>Anika</td>\n",
       "      <td>F</td>\n",
       "      <td>10</td>\n",
       "      <td>Osborne</td>\n",
       "      <td>227.99628</td>\n",
       "      <td>free</td>\n",
       "      <td>Philadelphia-Camden-Wilmington, PA-NJ-DE-MD</td>\n",
       "      <td>PUT</td>\n",
       "      <td>...</td>\n",
       "      <td>1262</td>\n",
       "      <td>I'm In Miami Bitch</td>\n",
       "      <td>200</td>\n",
       "      <td>1540478927000</td>\n",
       "      <td>\"Mozilla/5.0 (iPhone; CPU iPhone OS 7_1_2 like...</td>\n",
       "      <td>13</td>\n",
       "      <td>2018-10-25 14:48:47</td>\n",
       "      <td>2018-08-02 06:40:32</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           artist       auth firstName gender  itemInSession lastName  \\\n",
       "185153    Jamie T  Logged In    Daniel      M            161    Marks   \n",
       "12368   Energy 52  Logged In     Alexi      F             30   Warren   \n",
       "178657     P.O.D.  Logged In     Erick      M            153   Brooks   \n",
       "190882  DJ Khaled  Logged In   Kaylenn      F             82  Jenkins   \n",
       "256047      LMFAO  Logged In     Anika      F             10  Osborne   \n",
       "\n",
       "           length level                                     location method  \\\n",
       "185153  259.65669  paid           Chicago-Naperville-Elgin, IL-IN-WI    PUT   \n",
       "12368   462.13179  free                   Spokane-Spokane Valley, WA    PUT   \n",
       "178657  258.16771  paid                                    Selma, AL    PUT   \n",
       "190882  227.08200  paid                                    Price, UT    PUT   \n",
       "256047  227.99628  free  Philadelphia-Camden-Wilmington, PA-NJ-DE-MD    PUT   \n",
       "\n",
       "          ...    sessionId                                               song  \\\n",
       "185153    ...           94                                  Sheila (Explicit)   \n",
       "12368     ...          963                                    CafÃÂ© Del Mar   \n",
       "178657    ...         1069  Youth Of The Nation (2006 Remastered Album Ver...   \n",
       "190882    ...          649  All I Do Is Win (feat. T-Pain_ Ludacris_ Snoop...   \n",
       "256047    ...         1262                                 I'm In Miami Bitch   \n",
       "\n",
       "        status             ts  \\\n",
       "185153     200  1539972314000   \n",
       "12368      200  1539789477000   \n",
       "178657     200  1540023127000   \n",
       "190882     200  1540054668000   \n",
       "256047     200  1540478927000   \n",
       "\n",
       "                                                userAgent  userId  \\\n",
       "185153  \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_4...  100012   \n",
       "12368   Mozilla/5.0 (Windows NT 6.1; WOW64; rv:32.0) G...      54   \n",
       "178657  \"Mozilla/5.0 (Windows NT 6.3; WOW64) AppleWebK...      58   \n",
       "190882  \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_2...     141   \n",
       "256047  \"Mozilla/5.0 (iPhone; CPU iPhone OS 7_1_2 like...      13   \n",
       "\n",
       "                       time    registration_time churn downgrade  \n",
       "185153  2018-10-19 18:05:14  2018-09-19 18:19:14     1         1  \n",
       "12368   2018-10-17 15:17:57  2018-07-25 01:37:42     1         1  \n",
       "178657  2018-10-20 08:12:07  2018-09-26 10:12:31     1         0  \n",
       "190882  2018-10-20 16:57:48  2018-09-02 15:57:58     0         1  \n",
       "256047  2018-10-25 14:48:47  2018-08-02 06:40:32     0         1  \n",
       "\n",
       "[5 rows x 22 columns]"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_pandas.sample(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's observe the spark dataframe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+\n",
      "|userId|\n",
      "+------+\n",
      "|    10|\n",
      "|   100|\n",
      "|100001|\n",
      "|100002|\n",
      "|100003|\n",
      "+------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df.select('userId').dropDuplicates().sort('userId').show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+-----+---------+\n",
      "|userId|churn|downgrade|\n",
      "+------+-----+---------+\n",
      "|100010|    0|        0|\n",
      "|200002|    0|        0|\n",
      "|   125|    1|        0|\n",
      "|   124|    0|        0|\n",
      "|    51|    1|        0|\n",
      "|     7|    0|        0|\n",
      "|    15|    0|        0|\n",
      "|    54|    1|        1|\n",
      "|   155|    0|        0|\n",
      "|100014|    1|        0|\n",
      "|   132|    0|        0|\n",
      "|   154|    0|        0|\n",
      "|   101|    1|        0|\n",
      "|    11|    0|        1|\n",
      "|   138|    0|        0|\n",
      "|300017|    0|        0|\n",
      "|100021|    1|        0|\n",
      "|    29|    1|        0|\n",
      "|    69|    0|        0|\n",
      "|   112|    0|        0|\n",
      "+------+-----+---------+\n",
      "only showing top 20 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df.select(['userId', 'churn', 'downgrade']).dropDuplicates().show(20)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we have defined **churn** and **downgrade** events as seperate columns."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "### Explore Data\n",
    "Once you've defined churn, perform some exploratory data analysis to observe the behavior for users who stayed vs users who churned. You can start by exploring aggregates on these two groups of users, observing how much of a specific action they experienced per a certain time unit or number of songs played."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Gender distribution on churn event\n",
    "\n",
    "First, we can explore the gender distribution on churn event to understand gender has an effect or not.\n",
    "\n",
    "[1] https://medium.com/@andykashyap/top-5-tricks-to-make-plots-look-better-9f6e687c1e08"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-----+------+-----+\n",
      "|churn|gender|count|\n",
      "+-----+------+-----+\n",
      "|    0|     M|   89|\n",
      "|    0|     F|   84|\n",
      "|    1|     F|   20|\n",
      "|    1|     M|   32|\n",
      "+-----+------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df.dropDuplicates([\"userId\", \"gender\"]).groupby([\"churn\", \"gender\"]).count().sort(\"churn\").show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pandas_gender = df.dropDuplicates([\"userId\", \"gender\"]).groupby([\"churn\", \"gender\"]).count().sort(\"churn\").toPandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmUAAAH1CAYAAAC3AYn5AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAIABJREFUeJzt3Xu4ZWV9H/DvTy4OlwQFRqOMMGMlXEREGDEEpQRUvBDFRquGWAwqJl5CYo3RVhpNatWntGmCxkgCxVYLWFDhialRgTH1UiKIQS5REEacYGAkauQqo2//WHvG4/HMzDkz5+z9zpzP53nWc85619pr/dbee/b5zrvetXa11gIAwGQ9ZNIFAAAglAEAdEEoAwDogFAGANABoQwAoANCGQBAB4Qy2AZU1RuranUHdayqqvcswHaXV1WrqpWj+WNH83vP975G21+Q49gSVXVaVd1WVT+qqrdtxXYW9DkDFp5QBrNQVY+sqj+qqpuq6v6qurOqPl9Vr6+q3Sdd39aoqvNGf8xbVT04OrYrquq1VbXTtNX/VZK3zHK7b6uq62ZZxjeTPCrJl+dQ+mxqeHlV3T3Dolkfx0KqqocneW+S/5xknyRnbmLdw6rqwqr6x9F78ObRa/eEcdU7V1PC9kzTs8ZcSzdBHDZmx0kXAL2rquVJPpfkn5OckeTaDP+h+fkk/ybJXUn+14TKm7Wq2rm19oONLP50kpcl2SHJ0iTHJXl7kpdV1fGttXuSpLX2TwtY1z/O97Y3ZiGOYwvtl+Fz+C9ba9/a2EpVdWKSi/Pj1+nmJHsl+ZUk70ry3IUqsKoekqRaaz/cis08K8nfTWvr5TWAfrTWTCbTJqYk/ydDT85uG1leU37fI8nZSe5M8v0kn0mycsrylye5O8nxSa5Lck+SK5KsmLbNN2UIKXcn+R9J3pZk9bR1fj3JDUnuT/K1JL+T5CFTlrckr03ykdF+ztxI/edlCAXT2w9J8oMkb5/StirJe6bM/6sMIfW+DH9kP5PkkaPjbNOml2+sriTLR+0rR+scO5o/MUPv2f1Jrk5yxPTnclrN6x+395Tfp05v28hxPDzJB5J8Z3Qsn07y+Lm+bjM8h/sm+ejovfD90TEvm7LN6fUtn2EbuyZZm+TSjezjYdOO/fgkVya5N8lVSQ6f7XM27VifMzrWdaP3wnlJ/jLJ6Un+YfRc/fcku27i+H/idZ1h+c+Plj9hWvtpSb6dZKfR/MFJPj56Du9Mcn6Sn5v+Ht5YbaPlm32uTaZJT05fwiZU1Z5JTkjy3jbqLZqutdZG61aGPxz7ZAgTT0ryN0kur6pHTXnIQzOcOjs1yVFJHpbkz6bs818n+Y9Jfj/J4Um+muQN0+p6VZL/lOQ/JDkoyb9N8ntJXjOtvN9P8ldJnpDhNNmstdauS/KJDL0xP6Wqfi7JBRnCzEFJjknyP0eLL0zyX0a1P2o0XbgFdZ2Z4bhWJrklyceratdZHsLnk/x2hnCyvoaNnR48L8lTkjw/yZGjx3yiqnaZss4mX7fpRu+Hj2UIqccl+aUkj07ysdGyCzP0IGW0z0dlCP/TnZAhZL5rpv201r47remdSd6c4b1zV5IPjfY3F0uSvDXJqzMEom+M2p+WIaA9PcmLk7wgQxDaIq21r2UIjidPW3Rykgtbaw+O/u38TYaAeORo37snuXTUi7fepmo7PckXMgS19e+FmZ5rmKxJp0KTqecpwx/qluQF09rXZOhNuDvJn43ajhvN7zJt3S8nedPo95ePtnfAlOUnZ+iResho/vNJ/nzaNj6dKT1lSW5L8rJp6/x2khumzLckZ83iGM/LDD1lo2XvSnLvlPlVGfUwZfij35Lst5HHvi3JdTO0/1Rd2XhP2clT1tk9yXeTvHLKczmrXp8Zaph6HPuPHnPMlOV7JPnetH1t8nWbYR/PSPLDTOmRSfLYJD9K8vTR/MpsptcmQ69pS/LwzbyO64/9hCltR4/apvbOzeY5a5nSKznlffLNJDtOafvzJJ/eRE3rX9d78+N/L+unPUbrnJ4h9NVo/jGj5+io0fwfJLls2nYfPtrukbOtLdN6R02mHic9ZbBlnpbksCR/m6FXIUmOyOhUU1XdvX7K8L/3fzHlsQ+01r46Zf72JDtl6HlJhl6nL0zb34b5qlqa4Q/X+6ft513T9pMMvRBbozL88ZvJ32UIi9dV1cVV9Zuj2mZjtnVtOO7W2t1JvpKh52Y+HZQhBEzd1/dm2NfmXreZtnt7a231lO3eMnrcXI5hrr1c106rMUkeMcdtrMvMF13c0FpbN237s9n2r2b49zJ1+v5o2fkZehCfNmXdW1pr61+PI5IcM+29vr6Xa+r7fUtrg24Y6A+bdnOGUHLg1MbW2q1JUlX3Tml+SJI78uM/LlP985Tf101btj70zPY/SevX+40MvWqbMuMp1zk4OMNpw5/SWvthVT0zyS8keWaSVyR5Z1X9y9ba9EHd811XMgSp6YFl+tWis7Gp0DM1kM71ddtUoN1Y+0y+Nvp5UDb/eifJgzPsZ32Ns33OHmgzD+x/cNp8y+zet2taazfPtKC1dmdVfTpDz+PfjH5+aMoqD8kwLOCNMzz8jnmoDbrhDQub0Fq7K8knk7xuFre++FKG8UM/aq3dPG26cw67vTFD0Jlqw3xr7Y4Mg5n/xQz7mfEP35aoqkMyjHm6aGPrtMEXWmtvT/LkDL0TLx4t/kGGqzm3xobjrqrdMvQ63jhqWptk16r62SnrHzbt8bOp4YYMn4VHTdnXz2YY73bDlpW9Ybv7jK7eXb/dx2boFZrLdj+ZYdD7m2daWFUb66mbyWyes0n4YJIXVdURGZ73D05Z9qUkj0/yjRne79+faWMbMR/vR1hQQhls3msy/Fu5uqpeWlUHV9XPV9VLkzwxw7ihZDiV97kkl1TVs6tqRVUdVVVvr6qZes825o+TnFJVr6qq/avqLRnGtk31tiRvqqrfqaoDquqQqvo3o3W3xEOr6ueq6tFV9cSqekOGMThXZyOD46vqF6rqrVX15KraN8nzMpxWXR84VifZr6oOr6q9q+qhW1DXW6vqGVX1+CTnZvjDuv72I1dm6HF7Z1U9rqp+JT99ocPqJEtG29h7posEWms3Jbkkw+ngp43u+/XBDL2bW3Ork09nOMX7oao6ooYb434oQ8i4fLYbacMFJq9M8qyq+vjoWJaPntc/zE/2Km3ObJ6zhbDX6P01dZp6EcVHM/TYnZPkb0evyXrvzTDG78KqekpVPbaqnl5VZ1fVz8yhhtVJjhw9d3tPu0gAuuBNCZsxGgf0pAxXIv5hkmsy/GF9Q5I/zTDAPq21luE2ApdnGGT81SQfTnJAfjy2Zzb7uzBD6HrHaF9PSPJfp63zFxmuAnxZhj/8/zfDbQRu3aKDHK5Y+1aGCwguyxCw3p5h8PvGTjV+L8NA8r9MclOGqy3/sLW2vpfj4gxXWF6WoYfmpVtQ15tH2/1ShgH5J7afvGfayRkG1H8lw/GfMfXBrbXPZ7hC8vxRDW/ayH5+PcP4wEtHP3dN8qzW2n1bUPP6fbckJ432uyrDLTT+MclJo2Vz2dYlGXry7s0QGL+a5H9nCMEbO6aZtrPZ52yBfCLD+2vqtOGKy9bavRmC2RPzk71kaa3dnuF99qPRdq7PENQeGE2zdWaGUH9Dhtdk3y07FFg4NcfPBgAAFoCeMgCADghlAAAdEMoAADoglAEAdEAoAwDowDZ5R/+99967LV++fNJlAABs1tVXX/3t1tpmv4Zumwxly5cvz1VXbe1X+gEALLyq+sZs1nP6EgCgA0IZAEAHhDIAgA5sk2PKAIAfe/DBB7NmzZrcf//9ky5lUVuyZEmWLVuWnXbaaYseL5QBwDZuzZo1+Zmf+ZksX748VTXpchal1lruuuuurFmzJitWrNiibTh9CQDbuPvvvz977bWXQDZBVZW99tprq3orhTIA2A4IZJO3ta+BUAYAbLU77rgjv/qrv5rHPvaxOeKII3LUUUflox/96FZvd9WqVTnxxBPnocL+CWUAwFZpreWkk07KMccck1tuuSVXX311LrjggqxZs2bstaxbt27s+5wvQhkAsFUuv/zy7LzzzvmN3/iNDW377bdfXv/61+eHP/xhfvd3fzdPfvKTc+ihh+b9739/kqEH7Nhjj80LX/jCHHjggTn55JPTWkuSfOITn8iBBx6Ypz71qfnIRz6yYZv33HNPTj311Dz5yU/Ok570pFxyySVJkvPOOy8vetGL8su//Mt55jOfOcYjn1+uvgQAtsr111+fww8/fMZl55xzTvbYY4988YtfzAMPPJCjjz56Q3C65pprcv311+fRj350jj766Hzuc5/LypUr86pXvSqXX355Hve4x+XFL37xhm294x3vyHHHHZdzzz033/3ud3PkkUfm6U9/epLkC1/4Qq699trsueeeC3/AC0QoAwDm1Wtf+9p89rOfzc4775z99tsv1157bS666KIkyfe+973cdNNN2XnnnXPkkUdm2bJlSZLDDjssq1evzu67754VK1Zk//33T5L82q/9Ws4+++wkySc/+clceumlOfPMM5MMV53edtttSZJnPOMZ23QgS4QyAGArPf7xj8/FF1+8Yf69731vvv3tb2flypXZd999c9ZZZ+WEE074icesWrUqD33oQzfM77DDDhvGg23sKsbWWi6++OIccMABP9F+5ZVXZrfddpuvw5kYY8oAgK1y3HHH5f7778/73ve+DW333ntvkuSEE07I+973vjz44INJkq997Wu55557NrqtAw88MLfeemu+/vWvJ0nOP//8DctOOOGEnHXWWRvGnl1zzTXzfiyTJJQBAFulqvKxj30sn/nMZ7JixYoceeSROeWUU/Lud787r3zlK3PwwQfn8MMPzyGHHJJXv/rVm7xCcsmSJTn77LPz3Oc+N0996lOz3377bVh2xhln5MEHH8yhhx6aQw45JGecccY4Dm9san3a3JasXLmyXXXVVZMuAwC6cOONN+aggw6adBlk5teiqq5ura3c3GP1lAEAdEAoAwDogFAGANABt8SYheNe885Jl8AWuvxP3zLpEgBgVvSUAQB0QCgDAOiAUAYAbLWqyste9rIN8+vWrcvSpUtz4oknTrCq+bXDDjvksMMO2zCtXr16XrdvTBkAbGfmeyz0bMbn7rbbbrnuuuty3333ZZdddsmnPvWp7LPPPvNax1SvP3fVvG7vrFOP3ew6u+yyS7785S/P636n0lMGAMyLZz/72fn4xz+eZPh6pJe+9KUTrmjbIpQBAPPiJS95SS644ILcf//9ufbaa/OUpzxl0iXNq/vuu2/DqcsXvOAF8759py8BgHlx6KGHZvXq1Tn//PPznOc8Z9LlzLuFPn0plAEA8+Z5z3te3vjGN2bVqlW56667Jl3ONkUoAwDmzamnnpo99tgjT3jCE7Jq1apJl7NNMaYMAJg3y5Yty+mnnz7pMrZJesoAYDszia+Yu/vuu3+q7dhjj82xxx67IPubzS0s5ttMxzif9JQBAHRAKAMA6IBQBgDQAaEMALYDrbVJl7Dobe1rIJQBwDZuyZIlueuuuwSzCWqt5a677sqSJUu2eBuuvgSAbdyyZcuyZs2arF27dtKlLGpLlizJsmXLtvjxQhkAbON22mmnrFixYtJlsJWcvgQA6IBQBgDQAaEMAKADQhkAQAcM9Ge79vpzV026BLbCJL7bDmBS9JQBAHRAKAMA6IBQBgDQAaEMAKADQhkAQAeEMgCADghlAAAdEMoAADoglAEAdEAoAwDogFAGANABoQwAoANCGQBAB4QyAIAOjDWUVdXvVNX1VXVdVZ1fVUuqakVVXVlVN1XVhVW18zhrAgDowdhCWVXtk+S3kqxsrR2SZIckL0ny7iR/1FrbP8l3krxiXDUBAPRi3Kcvd0yyS1XtmGTXJN9KclySi0bLP5DkpDHXBAAwcWMLZa21f0hyZpLbMoSx7yW5Osl3W2vrRqutSbLPuGoCAOjFOE9fPjzJ85OsSPLoJLslefYMq7aNPP60qrqqqq5au3btwhUKADAB4zx9+fQkt7bW1rbWHkzykSS/mORho9OZSbIsye0zPbi1dnZrbWVrbeXSpUvHUzEAwJiMM5TdluQXqmrXqqokxye5IckVSV44WueUJJeMsSYAgC6Mc0zZlRkG9H8pyVdG+z47ye8leUNV3ZxkryTnjKsmAIBe7Lj5VeZPa+33k/z+tOZbkhw5zjoAAHrjjv4AAB0QygAAOiCUAQB0QCgDAOiAUAYA0AGhDACgA0IZAEAHhDIAgA4IZQAAHRDKAAA6IJQBAHRAKAMA6IBQBgDQAaEMAKADQhkAQAeEMgCADghlAAAdEMoAADoglAEAdEAoAwDogFAGANABoQwAoANCGQBAB4QyAIAOCGUAAB0QygAAOiCUAQB0QCgDAOiAUAYA0AGhDACgA0IZAEAHhDIAgA4IZQAAHRDKAAA6IJQBAHRAKAMA6IBQBgDQAaEMAKADQhkAQAeEMgCADghlAAAdEMoAADoglAEAdEAoAwDogFAGANABoQwAoANCGQBAB4QyAIAOCGUAAB0QygAAOiCUAQB0QCgDAOiAUAYA0AGhDACgA0IZAEAHhDIAgA4IZQAAHRDKAAA6IJQBAHRAKAMA6IBQBgDQAaEMAKADQhkAQAeEMgCADghlAAAdEMoAADoglAEAdEAoAwDogFAGANABoQwAoANCGQBAB4QyAIAOCGUAAB0QygAAOiCUAQB0QCgDAOiAUAYA0AGhDACgA0IZAEAHhDIAgA4IZQAAHRDKAAA6IJQBAHRAKAMA6MBYQ1lVPayqLqqqv6+qG6vqqKras6o+VVU3jX4+fJw1AQD0YNw9ZX+c5BOttQOTPDHJjUnenOSy1tr+SS4bzQMALCpjC2VV9bNJjklyTpK01n7QWvtukucn+cBotQ8kOWlcNQEA9GKcPWWPTbI2yX+vqmuq6i+qarckj2ytfStJRj8fMcaaAAC6MM5QtmOSw5O8r7X2pCT3ZA6nKqvqtKq6qqquWrt27ULVCAAwEeMMZWuSrGmtXTmavyhDSLujqh6VJKOfd8704Nba2a21la21lUuXLh1LwQAA4zK2UNZa+8ck36yqA0ZNxye5IcmlSU4ZtZ2S5JJx1QQA0Isdx7y/1yf5UFXtnOSWJL+eIRh+uKpekeS2JC8ac00AABM31lDWWvtykpUzLDp+nHUAAPTGHf0BADoglAEAdEAoAwDogFAGANABoQwAoANCGQBAB4QyAIAOCGUAAB0QygAAOiCUAQB0QCgDAOiAUAYA0AGhDACgA0IZAEAHhDIAgA4IZQAAHRDKAAA6IJQBAHRAKAMA6IBQBgDQAaEMAKADQhkAQAeEMgCADghlAAAdEMoAADoglAEAdEAoAwDogFAGANABoQwAoANCGQBAB4QyAIAOCGUAAB0QygAAOiCUAQB0QCgDAOiAUAYA0AGhDACgA7MOZVV1TFXtOEP7jlV1zPyWBQCwuMylp+yKJHvO0L7HaBkAAFtoLqGskrQZ2vdKcs/8lAMAsDj91OnI6arq0tGvLckHq+qBKYt3SHJIks8vQG0AAIvGZkNZkrtGPyvJd5LcN2XZD5J8Nsmfz3NdAACLymZDWWvt15OkqlYnObO15lQlAMA8m01PWZKktfb2hSwEAGAxm3Uoq6o9k7wjyfFJHpFpFwm01n52fksDAFg8Zh3KkpyT5ElJzk5ye2a+EhMAgC0wl1B2fJJntNauXKhiAAAWq7ncp+zOJHcvVCEAAIvZXELZv0/yB1W1+0IVAwCwWM3l9OVbkyxPcmdVfSPJg1MXttYOnce6AAAWlbmEsosWrAoAgEXOfcoAADowlzFlAAAskLncPPb72cS9ydw8FgBgy81lTNnrps3vlOFmsr+S4U7/AABsobmMKfvATO1V9aUMN5Y9a76KAgBYbOZjTNkVSX55HrYDALBozUcoe0mSb8/DdgAAFq25DPT/Sn5yoH8leWSSPZP85jzXBQCwqGzNzWN/lGRtklWttb+fv5IAABYfN48FAOjAXHrKkiRVdVySgzOcyry+tbZqvosCAFhs5jKmbJ8kH01yRJLbR82PrqqrkrygtXb7Rh8MAMAmzeXqyz9J8sMkj2utPaa19pgk+4/a/mQhigMAWCzmcvryGUmOba3dur6htXZLVf1WksvmvTIAgEVkPu5T9qN52AYAwKI2l1B2WZI/qarHrG+oqn2T/HH0lAEAbJW5hLLfSrJrkluq6htVtTrJ10dtv7UAtQEALBpzuU/ZN5McXlXPSHJghjv639Ba+/RCFQcAsFhstqesqp5dVaurao8kaa19qrV2VmvtT5J8cbTsmQteKQDAdmw2py9fl+Q/t9a+N33BqO3dSU6f78IAABaT2YSyQ5Ns6hTl5UmeOD/lAAAsTrMJZUuz6dtetCR7zU85AACL02xC2ZoMvWUbc2iSf5ifcgAAFqfZhLKPJ/nDqtpl+oKq2jXJH4zWAQBgC83mlhjvSPLCJDdV1VlJ/n7UflCGiwAqyX9amPIAABaHzYay1tqdVfWLSd6XIXzV+kVJ/jrJa1prdyxciQAA279Z3Ty2tfaNJM+pqocneVyGYHZTa+07C1kcAMBiMes7+ifJKIR9cYFqAQBYtOby3ZcAACwQoQwAoANCGQBAB4QyAIAOCGUAAB0Yeyirqh2q6pqq+svR/IqqurKqbqqqC6tq53HXBAAwaZPoKTs9yY1T5t+d5I9aa/sn+U6SV0ygJgCAiRprKKuqZUmem+QvRvOV5LgkF41W+UCSk8ZZEwBAD8bdU/bfkrwpyY9G83sl+W5rbd1ofk2SfcZcEwDAxI0tlFXViUnubK1dPbV5hlXbRh5/WlVdVVVXrV27dkFqBACYlHH2lB2d5HlVtTrJBRlOW/63JA+rqvVf97Qsye0zPbi1dnZrbWVrbeXSpUvHUS8AwNiMLZS11t7SWlvWWlue5CVJLm+tnZzkiiQvHK12SpJLxlUTAEAverhP2e8leUNV3ZxhjNk5E64HAGDsdtz8KvOvtbYqyarR77ckOXISdQAA9KKHnjIAgEVPKAMA6IBQBgDQAaEMAKADQhkAQAeEMgCADghlAAAdEMoAADoglAEAdEAoAwDogFAGANABoQwAoANCGQBAB4QyAIAOCGUAAB0QygAAOiCUAQB0QCgDAOiAUAYA0AGhDACgA0IZAEAHhDIAgA4IZQAAHRDKAAA6IJQBAHRAKAMA6IBQBgDQAaEMAKADQhkAQAeEMgCADghlAAAdEMoAADqw46QLAFisjnvNOyddAlvo8j99y6RLYDukpwwAoANCGQBAB4QyAIAOCGUAAB0QygAAOiCUAQB0QCgDAOiAUAYA0AGhDACgA0IZAEAHhDIAgA4IZQAAHRDKAAA6IJQBAHRAKAMA6IBQBgDQAaEMAKADQhkAQAeEMgCADghlAAAdEMoAADoglAEAdEAoAwDogFAGANABoQwAoANCGQBAB4QyAIAOCGUAAB0QygAAOiCUAQB0QCgDAOiAUAYA0AGhDACgA0IZAEAHhDIAgA4IZQAAHRDKAAA6IJQBAHRAKAMA6IBQBgDQAaEMAKADQhkAQAeEMgCADghlAAAdEMoAADoglAEAdEAoAwDogFAGANABoQwAoANCGQBAB4QyAIAOjC2UVdVjquqKqrqxqq6vqtNH7XtW1aeq6qbRz4ePqyYAgF6Ms6dsXZJ/21o7KMkvJHltVR2c5M1JLmut7Z/kstE8AMCiMrZQ1lr7VmvtS6Pfv5/kxiT7JHl+kg+MVvtAkpPGVRMAQC8mMqasqpYneVKSK5M8srX2rWQIbkkeMYmaAAAmacdx77Cqdk9ycZLfbq39c1XN9nGnJTktSfbdd9+FKxAANuP1566adAlshbNOPXbSJcxorD1lVbVThkD2odbaR0bNd1TVo0bLH5Xkzpke21o7u7W2srW2cunSpeMpGABgTMZ59WUlOSfJja21/zpl0aVJThn9fkqSS8ZVEwBAL8Z5+vLoJC9L8pWq+vKo7d8leVeSD1fVK5LcluRFY6wJAKALYwtlrbXPJtnYALLjx1UHAECP3NEfAKADQhkAQAeEMgCADghlAAAdEMoAADoglAEAdEAoAwDogFAGANABoQwAoANCGQBAB4QyAIAOCGUAAB0QygAAOiCUAQB0QCgDAOiAUAYA0AGhDACgA0IZAEAHhDIAgA4IZQAAHRDKAAA6IJQBAHRAKAMA6IBQBgDQAaEMAKADQhkAQAeEMgCADghlAAAdEMoAADoglAEAdEAoAwDogFAGANABoQwAoANCGQBAB4QyAIAOCGUAAB0QygAAOiCUAQB0QCgDAOiAUAYA0AGhDACgA0IZAEAHhDIAgA4IZQAAHRDKAAA6IJQBAHRAKAMA6IBQBgDQAaEMAKADQhkAQAeEMgCADghlAAAdEMoAADoglAEAdEAoAwDogFAGANABoQwAoANCGQBAB4QyAIAOCGUAAB0QygAAOiCUAQB0QCgDAOiAUAYA0AGhDACgA0IZAEAHhDIAgA4IZQAAHRDKAAA6IJQBAHRAKAMA6IBQBgDQAaEMAKADQhkAQAeEMgCADghlAAAdEMoAADoglAEAdEAoAwDogFAGANABoQwAoANCGQBAB4QyAIAOCGUAAB3oIpRV1bOq6qtVdXNVvXnS9QAAjNvEQ1lV7ZDkvUmeneTgJC+tqoMnWxUAwHhNPJQlOTLJza21W1prP0hyQZLnT7gmAICx6iGU7ZPkm1Pm14zaAAAWjR0nXUCSmqGt/dRKVaclOW00e3dVfXVBq2K7cEWyd5JvT7oOtsx7XjHpCmBmPlu2bRP4bNlvNiv1EMrWJHnMlPllSW6fvlJr7ewkZ4+rKLYPVXVVa23lpOsAti8+W1gIPZy+/GKS/atqRVXtnOQlSS6dcE0AAGM18Z6y1tq6qnpdkr9OskOSc1tr10+4LACAsZp4KEuS1tpfJfmrSdfBdskpb2Ah+Gxh3lVrPzWmHgCAMethTBkAwKInlLHd8vVdwHyrqnOr6s6qum7StbD9EcrYLvn6LmCBnJfkWZMugu2TUMb2ytd3AfOutfY3Sf5p0nWwfRLK2F75+i4AtilCGdsVkYr3AAACo0lEQVSrWX19FwD0QihjezWrr+8CgF4IZWyvfH0XANsUoYztUmttXZL1X991Y5IP+/ouYGtV1flJvpDkgKpaU1WvmHRNbD/c0R8AoAN6ygAAOiCUAQB0QCgDAOiAUAYA0AGhDACgA0IZsF2rqlZVL5x0HQCbI5QB27SqemRV/XFVfb2qHqiqf6iq/1NVz5l0bQBzseOkCwDYUlW1PMnnknw/yVuS/F2G/2wen+TPkuy7gPveqbX24EJtH1h89JQB27I/zfDl8ytbax9urX21tXZja+09SZ44Zb09q+p/V9U9VXVLVf3a+gVVtXx0inPl1A1PPe05ZZ2XVtXlVXVfkldX1cur6u6qOr6qrhtt/4qqWjGGYwe2M0IZsE2qqj2TPCvJe1prd09f3lr7zpTZ/5DkkgxB7cIk51bVfluw23dmCIIHJ/nYqO2hGXrpTk1yVJKHZeilA5gToQzYVj0uQy/ZjbNY93+21j7YWrs5yRlJ1iV52hbs86zW2kWttVtba2tGbTsmeW1r7W9ba9cmOTPJL1WVz1dgTnxoANuqmsO6167/ZfRl9WuTPGIL9nnVDG0PtNa+OmX+9iQ7ZegxA5g1oQzYVt2UpCU5aBbrTh+Q3/Ljz78fjX5uCHlVtdNGtnPPDG3rZth24vMVmCMfGsA2qbX2T0n+Osnrqmr36curarY9VWtHPx81pe2wrSwPYM6EMmBb9poMPVxXVdWLquqAqjqwqn4zU05Zbkpr7b4k/y/J71XV46vqFzOMCwMYK6EM2Ga11m5NcniSTyV5d4YgdnmS5yV59Rw2dero5xeTvD/JW+exTIBZqdba5tcCAGBB6SkDAOiAUAYA0AGhDACgA0IZAEAHhDIAgA4IZQAAHRDKAAA6IJQBAHRAKAMA6MD/B42OYq0B3jv2AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f45068adcc0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,8))\n",
    "ax = sns.barplot(x = 'churn', y = 'count', hue = 'gender', data = df_pandas_gender,  palette = \"Blues_d\")\n",
    "plt.xlabel('Churn', fontsize = 14)\n",
    "plt.ylabel('Count', fontsize = 14)\n",
    "plt.legend(title = 'Gender', loc = 'best', ncol = 2)\n",
    "\n",
    "plot = plt.title('Gender Distribution of Churn Event', fontsize = 14)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It is seen that male customers are slightly like to be churn than the female users. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Number of played/listened songs\n",
    "\n",
    "It might be a good indicator to understand the number of played songs in a session among the genders. Therefore, we may see the reason of male users have more likely to be churned.\n",
    "\n",
    "[1] https://www.datacamp.com/community/blog/seaborn-cheat-sheet-python\n",
    "\n",
    "[2] https://seaborn.pydata.org/generated/seaborn.violinplot.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pandas_song = df.where('page == \"NextSong\"').groupby(['churn', 'userId', 'sessionId','gender']).count()\\\n",
    "    .groupby(['churn', 'userId', 'gender']).agg({'count':'avg'})\\\n",
    "    .withColumnRenamed('avg(count)', 'avg_songs_played')\\\n",
    "    .toPandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f44e7548320>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,8))\n",
    "ax = sns.violinplot('churn', y = 'avg_songs_played', hue = 'gender', data = df_pandas_song, palette=\"Set2\", split = True, inner=\"quartile\")\n",
    "plt.xlabel('churn', fontsize = 12)\n",
    "plt.ylabel('Average count of listened songs per session', fontsize = 12)\n",
    "plt.legend(title = 'Gender', loc = 'best', ncol = 2)\n",
    "plt.title('Listened music distribution among gender')\n",
    "sns.despine(ax=ax);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It is seen that churned users had listened less music than the others. And also churned female users average listened songs are more than males. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Notes on Violin Plot\n",
    "\n",
    "A violin plot plays a similar role as a box and whisker plot. It shows the distribution of quantitative data across several levels of one (or more) categorical variables such that those distributions can be compared. Unlike a box plot, in which all of the plot components correspond to actual datapoints, the violin plot features a kernel density estimation of the underlying distribution.\n",
    "\n",
    "Violin plots have many of the same summary statistics as box plots:\n",
    "* The white dot represents the median\n",
    "* The thick gray bar in the center represents the interquartile range\n",
    "* The thin gray line represents the rest of the distribution, except for points that are determined to be “outliers” using a method that is a function of the interquartile range.\n",
    "\n",
    "On each side of the gray line is a kernel density estimation to show the distribution shape of the data. Wider sections of the violin plot represent a higher probability that members of the population will take on the given value; the skinnier sections represent a lower probability.\n",
    "\n",
    "[1] https://mode.com/blog/violin-plot-examples\n",
    "\n",
    "[2] https://blog.bioturing.com/2018/05/16/5-reasons-you-should-use-a-violin-graph/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pandas_likes = df.where('page == \"NextSong\" OR page == \"Thumbs Up\"').groupby(['userId', 'churn', 'gender', 'page']).count().toPandas()\n",
    "df_pandas_likes = df_pandas_likes.pivot_table(index=['userId','churn','gender'], values='count', columns='page').reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f44e3940278>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,8))\n",
    "ax = sns.violinplot(data = df_pandas_likes, x = 'churn', y = 'NextSong', hue = 'gender', split = True,  inner = \"quartile\", palette=\"Set2\")\n",
    "plt.xlabel('Churn Event')\n",
    "plt.ylabel('Played songs count according to NextSong Event')\n",
    "plt.legend(title = 'Gender', loc = 'best')\n",
    "title = plt.title('Total listening song distribution effect on churn by gender')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It seems that the mean of the total listened songs do not change that much among churned and unchurned users. However, we can say that female users have larger mean value and distribution than the male users."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f44e445c748>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,8))\n",
    "ax = sns.violinplot(data = df_pandas_likes, x = 'churn', y = 'Thumbs Up', hue = 'gender', split = True,  inner = \"quartile\", palette=\"Set3\")\n",
    "plt.xlabel('Churn event')\n",
    "plt.ylabel('Liked song count according to ThumbsUp event')\n",
    "plt.legend(title='Gender', loc='best')\n",
    "title = plt.title('Total liked song distribution effect on churn by gender')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Also for the liked song distribution, churned and normal users have similar characteristics."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Churned user status\n",
    "\n",
    "It might me a good indicator to see which payment status of user had churned most."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pandas_level = df.filter('page == \"Cancellation Confirmation\"').groupby('level').count().toPandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f44e8000278>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,8))\n",
    "ax = sns.barplot(data = df_pandas_level, x = 'level', y = 'count', palette=\"Set1\")\n",
    "plt.xlabel('Payment level', fontsize = 12)\n",
    "plt.ylabel('')\n",
    "plt.title('Churned user payment level distribution', fontsize = 14)\n",
    "sns.despine(ax=ax);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It is seen that paid users had churned more than free users."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Page distribution\n",
    "\n",
    "Let see the page distribution of churned and current users to understand the differences between them."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "churned_user = df_pandas[df_pandas.churn == 1].groupby(['page'])['userId'].count()\n",
    "churned_user = churned_user /churned_user.sum()*100\n",
    "\n",
    "active_user = df_pandas[df_pandas.churn == 0].groupby(['page'])['userId'].count()\n",
    "active_user = active_user /active_user.sum()*100\n",
    "\n",
    "users_df = pd.DataFrame({'Cancelled': churned_user,'Active users':active_user})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f4504eaaa90>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = users_df.plot(kind='barh', figsize=(12,10), cmap = \"plasma\");\n",
    "ax.set_xlabel('Percent of envent occurence (%)')\n",
    "ax.set_title('Percent of envent occurence for active and cancelled users');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Since Next Song creates a bias in the figure, I have decided to drop it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "churned_user = df_pandas[df_pandas.churn == 1].groupby(['page'])['userId'].count().drop('NextSong')\n",
    "# Define percentage to visualize better\n",
    "churned_user = churned_user /churned_user.sum()*100\n",
    "\n",
    "active_user = df_pandas[df_pandas.churn == 0].groupby(['page'])['userId'].count().drop('NextSong')\n",
    "# Define percentage to visualize better\n",
    "active_user = active_user /active_user.sum()*100\n",
    "\n",
    "users_df = pd.DataFrame({'Cancelled': churned_user,'Active users':active_user})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f450ff171d0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = users_df.plot(kind='barh', figsize=(12,10), cmap = \"plasma\");\n",
    "ax.set_xlabel('Percentage of envent occurence (%)')\n",
    "ax.set_title('Percentage distribution of page events among the active and cancelled users ');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Membership duration\n",
    "\n",
    "Lastly, we can investigate the membership duration of users to see effect on churn event."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pandas_lifetime = df\\\n",
    "    .select('userId','registration','ts','churn') \\\n",
    "    .withColumn('lifetime',(df.ts-df.registration)) \\\n",
    "    .groupBy('userId','churn') \\\n",
    "    .agg({'lifetime':'max'}) \\\n",
    "    .withColumnRenamed('max(lifetime)','lifetime') \\\n",
    "    .select('userId', 'churn', (col('lifetime')/1000/3600/24).alias('lifetime')) \\\n",
    "    .toPandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f45060845c0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = sns.boxplot(data = df_pandas_lifetime, y = 'churn', x = 'lifetime', orient = 'h', palette = \"Set2\")\n",
    "plt.xlabel('Days since registration when churn')\n",
    "plt.ylabel('Has customer churned')\n",
    "title = plt.title('Churned customer use the service for a shorter period of time')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Feature Engineering\n",
    "Once you've familiarized yourself with the data, build out the features you find promising to train your model on. To work with the full dataset, you can follow the following steps.\n",
    "- Write a script to extract the necessary features from the smaller subset of data\n",
    "- Ensure that your script is scalable, using the best practices discussed in Lesson 3\n",
    "- Try your script on the full data set, debugging your script if necessary\n",
    "\n",
    "If you are working in the classroom workspace, you can just extract features based on the small subset of data contained here. Be sure to transfer over this work to the larger dataset when you work on your Spark cluster."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Dataset Features:\n",
    "In the dataset, we have below features:\n",
    "\n",
    "<u>Categorical features</u>\n",
    "* Currently listening artists\n",
    "* Authorization\n",
    "* First/Last name of user\n",
    "* Gender of user\n",
    "* The payment level of membership (Paid/Free)\n",
    "* Location\n",
    "* Method\n",
    "* Current page\n",
    "* Currently listening song\n",
    "* User agent, (MacOs, Windows, etc.)\n",
    "\n",
    "<u>Numerical features</u>\n",
    "* Item in session\n",
    "* Length\n",
    "* Registration date in timestamp format\n",
    "* SessionId\n",
    "* Status\n",
    "* Time as ts in timestamp format\n",
    "* Userid\n",
    "\n",
    "\n",
    "For the analysis, we will add the below features according to the data exploration that we have performed in the previous section.\n",
    "\n",
    "1. Gender feature as numerical/binary variable\n",
    "2. Number of played/listened songs\n",
    "3. Number of liked songs according to thumbs up/down\n",
    "4. Number of songs added to playlist\n",
    "5. Number of songs listened per session\n",
    "6. Number of listened artists\n",
    "7. Number of added friends.\n",
    "8. Total time duration/lenght of listening event\n",
    "9. Membership duration\n",
    "10. Churn label\n",
    "11. Downgrade label\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+------+\n",
      "|userId|gender|\n",
      "+------+------+\n",
      "|    44|     1|\n",
      "|    46|     1|\n",
      "|    41|     1|\n",
      "|    72|     1|\n",
      "|300023|     1|\n",
      "+------+------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Gender feature as numerical/binary variable\n",
    "feature_1 = df.select(\"userId\", \"gender\").dropDuplicates().replace(['M', 'F'], ['0', '1'], 'gender').select('userId', col('gender').cast('int'))\n",
    "feature_1.show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+-----------+\n",
      "|userID|total_songs|\n",
      "+------+-----------+\n",
      "|100010|        381|\n",
      "|200002|        474|\n",
      "|   125|         11|\n",
      "|    51|       2464|\n",
      "|   124|       4825|\n",
      "+------+-----------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Number of played/listened songs\n",
    "feature_2 = df.select('userID','song').groupBy('userID').count().withColumnRenamed('count', 'total_songs')\n",
    "feature_2.show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+---------+\n",
      "|userID|thumbs_up|\n",
      "+------+---------+\n",
      "|100010|       17|\n",
      "|200002|       21|\n",
      "|    51|      100|\n",
      "|   124|      171|\n",
      "|     7|        7|\n",
      "+------+---------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Number of liked songs according to thumbs up/down\n",
    "feature_3 = df.select('userID','page').where(df.page == 'Thumbs Up').groupBy('userID').count().withColumnRenamed('count', 'thumbs_up')\n",
    "feature_3.show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+-----------+\n",
      "|userID|thumbs_down|\n",
      "+------+-----------+\n",
      "|100010|          5|\n",
      "|200002|          6|\n",
      "|    51|         21|\n",
      "|   124|         41|\n",
      "|     7|          1|\n",
      "+------+-----------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Number of liked songs according to thumbs up/down\n",
    "feature_4 = df.select('userID','page').where(df.page == 'Thumbs Down').groupBy('userID').count().withColumnRenamed('count', 'thumbs_down')\n",
    "feature_4.show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+---------------+\n",
      "|userID|add_to_playlist|\n",
      "+------+---------------+\n",
      "|100010|              7|\n",
      "|200002|              8|\n",
      "|    51|             52|\n",
      "|   124|            118|\n",
      "|     7|              5|\n",
      "+------+---------------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Number of songs added to playlist\n",
    "feature_5 = df.select('userID','page').where(df.page == 'Add to Playlist').groupBy('userID').count().withColumnRenamed('count', 'add_to_playlist')\n",
    "feature_5.show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+------------------+-----------------+\n",
      "|summary|            userId| avg_songs_played|\n",
      "+-------+------------------+-----------------+\n",
      "|  count|               225|              225|\n",
      "|   mean|65391.013333333336|70.78971233958933|\n",
      "| stddev|105396.47791907164| 42.6153697543817|\n",
      "|    min|                10|              3.0|\n",
      "|    max|                99|286.6666666666667|\n",
      "+-------+------------------+-----------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Number of songs listened per session\n",
    "feature_6 = df.where('page == \"NextSong\"').groupby(['userId', 'sessionId']).count().groupby(['userId']).agg({'count':'avg'}).withColumnRenamed('avg(count)', 'avg_songs_played')\n",
    "feature_6.describe().show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+------------+\n",
      "|userId|artist_count|\n",
      "+------+------------+\n",
      "|100010|         252|\n",
      "|200002|         339|\n",
      "|   125|           8|\n",
      "|    51|        1385|\n",
      "|   124|        2232|\n",
      "+------+------------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Number of listened artists\n",
    "feature_7 = df.filter(df.page==\"NextSong\").select(\"userId\", \"artist\").dropDuplicates().groupby(\"userId\").count().withColumnRenamed(\"count\", \"artist_count\")\n",
    "feature_7.show(5)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+----------+\n",
      "|userID|add_friend|\n",
      "+------+----------+\n",
      "|100010|         4|\n",
      "|200002|         4|\n",
      "|    51|        28|\n",
      "|   124|        74|\n",
      "|     7|         1|\n",
      "+------+----------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Number of added friends\n",
    "feature_8 = df.select('userID','page').where(df.page == 'Add Friend').groupBy('userID').count().withColumnRenamed('count', 'add_friend') \n",
    "feature_8.show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+------------------+\n",
      "|userID|       listen_time|\n",
      "+------+------------------+\n",
      "|100010| 66940.89735000003|\n",
      "|200002| 94008.87593999993|\n",
      "|   125|2089.1131000000005|\n",
      "|    51| 523275.8428000004|\n",
      "|   124|1012312.0927899999|\n",
      "+------+------------------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Total time duration/lenght of listening event\n",
    "feature_9 = df.select('userID','length').groupBy('userID').sum().withColumnRenamed('sum(length)', 'listen_time')\n",
    "feature_9.show(5)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+------------------+\n",
      "|userId|          lifetime|\n",
      "+------+------------------+\n",
      "|100010|  55.6436574074074|\n",
      "|200002| 70.07462962962963|\n",
      "|   125| 71.31688657407408|\n",
      "|    51|19.455844907407407|\n",
      "|   124|131.55591435185184|\n",
      "+------+------------------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Membership duration\n",
    "feature_10 = df.select('userId','registration','ts').withColumn('lifetime',(df.ts-df.registration)).groupBy('userId') \\\n",
    "    .agg({'lifetime':'max'}).withColumnRenamed('max(lifetime)','lifetime').select('userId', (col('lifetime')/1000/3600/24).alias('lifetime'))\n",
    "feature_10.show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+-----+\n",
      "|userId|label|\n",
      "+------+-----+\n",
      "|100010|    0|\n",
      "|200002|    0|\n",
      "|   125|    1|\n",
      "|   124|    0|\n",
      "|    51|    1|\n",
      "+------+-----+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Churn label\n",
    "churn_label = df.select('userId', col('churn').alias('label')).dropDuplicates()\n",
    "churn_label.show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+---------------+\n",
      "|userId|downgrade_label|\n",
      "+------+---------------+\n",
      "|100010|              0|\n",
      "|200002|              0|\n",
      "|   125|              0|\n",
      "|   124|              0|\n",
      "|    51|              0|\n",
      "+------+---------------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Downgrade label\n",
    "downgrade_label = df.select('userId', col('downgrade').alias('downgrade_label')).dropDuplicates()\n",
    "downgrade_label.show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------+-----------+---------+-----------+---------------+------------------+------------+----------+------------------+------------------+-----+---------------+\n",
      "|gender|total_songs|thumbs_up|thumbs_down|add_to_playlist|  avg_songs_played|artist_count|add_friend|       listen_time|          lifetime|label|downgrade_label|\n",
      "+------+-----------+---------+-----------+---------------+------------------+------------+----------+------------------+------------------+-----+---------------+\n",
      "|     1|        381|       17|          5|              7|39.285714285714285|         252|         4| 66940.89735000003|  55.6436574074074|    0|              0|\n",
      "|     0|        474|       21|          6|              8|              64.5|         339|         4| 94008.87593999993| 70.07462962962963|    0|              0|\n",
      "|     0|         11|        0|          0|              0|               8.0|           8|         0|2089.1131000000005| 71.31688657407408|    1|              0|\n",
      "|     1|       4825|      171|         41|            118|145.67857142857142|        2232|        74|1012312.0927899999|131.55591435185184|    0|              0|\n",
      "|     0|       2464|      100|         21|             52|             211.1|        1385|        28| 523275.8428000004|19.455844907407407|    1|              0|\n",
      "+------+-----------+---------+-----------+---------------+------------------+------------+----------+------------------+------------------+-----+---------------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "data = feature_1.join(feature_2,'userID','outer') \\\n",
    "    .join(feature_3,'userID','outer') \\\n",
    "    .join(feature_4,'userID','outer') \\\n",
    "    .join(feature_5,'userID','outer') \\\n",
    "    .join(feature_6,'userID','outer') \\\n",
    "    .join(feature_7,'userID','outer') \\\n",
    "    .join(feature_8,'userID','outer') \\\n",
    "    .join(feature_9,'userID','outer') \\\n",
    "    .join(feature_10,'userID','outer') \\\n",
    "    .join(churn_label,'userID','outer') \\\n",
    "    .join(downgrade_label,'userID','outer') \\\n",
    "    .drop('userID') \\\n",
    "    .fillna(0)\n",
    "\n",
    "data.show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Saving as SPARK format\n",
    "data.write.save('SparkFile_10.CSV', format='csv', header=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "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>gender</th>\n",
       "      <th>total_songs</th>\n",
       "      <th>thumbs_up</th>\n",
       "      <th>thumbs_down</th>\n",
       "      <th>add_to_playlist</th>\n",
       "      <th>avg_songs_played</th>\n",
       "      <th>artist_count</th>\n",
       "      <th>add_friend</th>\n",
       "      <th>listen_time</th>\n",
       "      <th>lifetime</th>\n",
       "      <th>label</th>\n",
       "      <th>downgrade_label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>381</td>\n",
       "      <td>17</td>\n",
       "      <td>5</td>\n",
       "      <td>7</td>\n",
       "      <td>39.285714</td>\n",
       "      <td>252</td>\n",
       "      <td>4</td>\n",
       "      <td>6.694090e+04</td>\n",
       "      <td>55.643657</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>474</td>\n",
       "      <td>21</td>\n",
       "      <td>6</td>\n",
       "      <td>8</td>\n",
       "      <td>64.500000</td>\n",
       "      <td>339</td>\n",
       "      <td>4</td>\n",
       "      <td>9.400888e+04</td>\n",
       "      <td>70.074630</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>11</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>8.000000</td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "      <td>2.089113e+03</td>\n",
       "      <td>71.316887</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>4825</td>\n",
       "      <td>171</td>\n",
       "      <td>41</td>\n",
       "      <td>118</td>\n",
       "      <td>145.678571</td>\n",
       "      <td>2232</td>\n",
       "      <td>74</td>\n",
       "      <td>1.012312e+06</td>\n",
       "      <td>131.555914</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>2464</td>\n",
       "      <td>100</td>\n",
       "      <td>21</td>\n",
       "      <td>52</td>\n",
       "      <td>211.100000</td>\n",
       "      <td>1385</td>\n",
       "      <td>28</td>\n",
       "      <td>5.232758e+05</td>\n",
       "      <td>19.455845</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0</td>\n",
       "      <td>201</td>\n",
       "      <td>7</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>21.428571</td>\n",
       "      <td>142</td>\n",
       "      <td>1</td>\n",
       "      <td>3.803409e+04</td>\n",
       "      <td>72.778183</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0</td>\n",
       "      <td>2278</td>\n",
       "      <td>81</td>\n",
       "      <td>14</td>\n",
       "      <td>59</td>\n",
       "      <td>136.714286</td>\n",
       "      <td>1302</td>\n",
       "      <td>31</td>\n",
       "      <td>4.773076e+05</td>\n",
       "      <td>56.513576</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1</td>\n",
       "      <td>3437</td>\n",
       "      <td>163</td>\n",
       "      <td>29</td>\n",
       "      <td>72</td>\n",
       "      <td>81.171429</td>\n",
       "      <td>1744</td>\n",
       "      <td>33</td>\n",
       "      <td>7.113449e+05</td>\n",
       "      <td>110.751690</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1</td>\n",
       "      <td>1002</td>\n",
       "      <td>58</td>\n",
       "      <td>3</td>\n",
       "      <td>24</td>\n",
       "      <td>136.666667</td>\n",
       "      <td>643</td>\n",
       "      <td>11</td>\n",
       "      <td>1.987793e+05</td>\n",
       "      <td>23.556019</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0</td>\n",
       "      <td>310</td>\n",
       "      <td>17</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>42.833333</td>\n",
       "      <td>233</td>\n",
       "      <td>6</td>\n",
       "      <td>6.770347e+04</td>\n",
       "      <td>85.083403</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   gender  total_songs  thumbs_up  thumbs_down  add_to_playlist  \\\n",
       "0       1          381         17            5                7   \n",
       "1       0          474         21            6                8   \n",
       "2       0           11          0            0                0   \n",
       "3       1         4825        171           41              118   \n",
       "4       0         2464        100           21               52   \n",
       "5       0          201          7            1                5   \n",
       "6       0         2278         81           14               59   \n",
       "7       1         3437        163           29               72   \n",
       "8       1         1002         58            3               24   \n",
       "9       0          310         17            3                7   \n",
       "\n",
       "   avg_songs_played  artist_count  add_friend   listen_time    lifetime  \\\n",
       "0         39.285714           252           4  6.694090e+04   55.643657   \n",
       "1         64.500000           339           4  9.400888e+04   70.074630   \n",
       "2          8.000000             8           0  2.089113e+03   71.316887   \n",
       "3        145.678571          2232          74  1.012312e+06  131.555914   \n",
       "4        211.100000          1385          28  5.232758e+05   19.455845   \n",
       "5         21.428571           142           1  3.803409e+04   72.778183   \n",
       "6        136.714286          1302          31  4.773076e+05   56.513576   \n",
       "7         81.171429          1744          33  7.113449e+05  110.751690   \n",
       "8        136.666667           643          11  1.987793e+05   23.556019   \n",
       "9         42.833333           233           6  6.770347e+04   85.083403   \n",
       "\n",
       "   label  downgrade_label  \n",
       "0      0                0  \n",
       "1      0                0  \n",
       "2      1                0  \n",
       "3      0                0  \n",
       "4      1                0  \n",
       "5      0                0  \n",
       "6      0                0  \n",
       "7      1                1  \n",
       "8      0                0  \n",
       "9      1                0  "
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_pandas = data.toPandas()\n",
    "data_pandas.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_pandas.to_csv('PandasFile_71.CSV', index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Modeling\n",
    "Split the full dataset into train, test, and validation sets. Test out several of the machine learning methods you learned. Evaluate the accuracy of the various models, tuning parameters as necessary. Determine your winning model based on test accuracy and report results on the validation set. Since the churned users are a fairly small subset, I suggest using F1 score as the metric to optimize. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Notes on F1 Score\n",
    "\n",
    "According to Wikipedia, F1 Score is defined as a weighted average of the precision and recall where an F1 score reached its best value at 1 and worst at 0. But first lets define the precision and recall:\n",
    "\n",
    "* The precision is the number of correct positive results divided by the number of all positive results. Precision talks about how precise/accurate your model is out of those predicted positive, how many of them are actual positive. Precision is a good measure to determine, when the costs of False Positive is high. For instance, email spam detection. In email spam detection, a false positive means that an email that is non-spam (actual negative) has been identified as spam (predicted spam). The email user might lose important emails if the precision is not high for the spam detection model.\n",
    "\n",
    "* The recall is the number of correct positive results divided by the number of positive results that should have been returned. Recall actually calculates how many of the Actual Positives our model capture through labeling it as Positive (True Positive). Applying the same understanding, we know that Recall shall be the model metric we use to select our best model when there is a high cost associated with False Negative.\n",
    "\n",
    "\n",
    "And accuracy is one of the more obvious metrics, it is the measure of all the correctly identified cases. It is most used when all the classes are equally important.\n",
    "\n",
    "F1 Score is defined as the harmonic mean of Precision and Recall and gives a better measure of the incorrectly classified cases than the Accuracy Metric. And the use case differences of accuracy and F1 Score can be summarized as below:\n",
    "\n",
    "1. Accuracy is used when the True Positives and True negatives are more important while F1-score is used when the False Negatives and False Positives are crucial\n",
    "\n",
    "2. Accuracy can be used when the class distribution is similar while F1-score is a better metric when there are imbalanced classes as in the above case.\n",
    "\n",
    "3. In most real-life classification problems, imbalanced class distribution exists and thus F1-score is a better metric to evaluate our model on.\n",
    "\n",
    "\n",
    "[1] https://towardsdatascience.com/accuracy-precision-recall-or-f1-331fb37c5cb9\n",
    "\n",
    "[2] https://adamyedidia.files.wordpress.com/2014/11/f_score.pdf\n",
    "\n",
    "[3] https://medium.com/analytics-vidhya/accuracy-vs-f1-score-6258237beca2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['gender', 'total_songs', 'thumbs_up', 'thumbs_down', 'add_to_playlist',\n",
       "       'avg_songs_played', 'artist_count', 'add_friend', 'listen_time',\n",
       "       'lifetime', 'label', 'downgrade_label'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_pandas.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will use the saved features as the inputs except the churn event, indicated as label. Because the churn label will be the feature that we want to predict by using the related features.\n",
    "\n",
    "After defining the features, we will create the feature vector and perform transformation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "# The input features are all features except the Churn Label and the output is the churn event.\n",
    "\n",
    "features_labels = ['gender', 'total_songs', 'thumbs_up', 'thumbs_down', 'add_to_playlist',\n",
    "                   'avg_songs_played', 'artist_count', 'add_friend', 'listen_time', 'lifetime', 'downgrade_label' ]\n",
    "\n",
    "# Define the vector assembler for all input columns\n",
    "features_vector  = VectorAssembler(inputCols = features_labels, outputCol = \"features\")\n",
    "\n",
    "input_data = features_vector.transform(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As a next step, we will perform standardization. Standardization is performed because using these variables without standardization in effect gives the variable with the larger range a weight of 100 in the analysis. Typical data standardization procedures equalize the range and/or data variability. Therefore we standardize the features by taking off the mean and divided by the standard deviation of each feature.\n",
    "\n",
    "[1] https://spark.apache.org/docs/latest/ml-features\n",
    "\n",
    "[2] https://scikit-learn.org/stable/modules/preprocessing.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "# standard scaler\n",
    "features_scaler = StandardScaler(inputCol = \"features\", outputCol=\"scaled_features\", withMean=True)\n",
    "features_scaler_fit = features_scaler.fit(input_data)\n",
    "scaled_inputs = features_scaler_fit.transform(input_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "After standardize our features, we can divide our dataset in to test, train, and validation. We can use randomSplit function to divide our dataset. \n",
    "\n",
    "RandomSplit is performing the randomly splits this DataFrame with the provided weights. When we perform random splitting we are splitting the data randomly to generate two sets: one to use during training of the ML algorithm (training set), and the second to check whether the training is working (test set). This is widely done and a very good idea, as it catches overfitting which otherwise can make it seem like you have a great ML solution when it's actually effectively just memorized the answer for each data point and can't interpolate or generalize.\n",
    "\n",
    "And one of the main problem is selecting the seed which is defined according to reference 4.\n",
    "\n",
    "[1] https://spark.apache.org/docs/2.2.0/api/python/pyspark.sql.html\n",
    "\n",
    "[2] https://stackoverflow.com/questions/40606456/randomsplit-dont-respect-specific-weights-pyspark\n",
    "\n",
    "[3] https://stackoverflow.com/questions/24857650/issue-understanding-splitting-data-in-scala-using-randomsplit-for-machine-lear\n",
    "\n",
    "[4] https://medium.com/@ODSC/properly-setting-the-random-seed-in-ml-experiments-not-as-simple-as-you-might-imagine-219969c84752"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Creating train, test, and validation\n",
    "train, rest = scaled_inputs.randomSplit([0.8, 0.2], seed = 45)\n",
    "validation, test = rest.randomSplit([0.5, 0.5], seed = 45)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Use baseline model to observe accuracy and F1-Score."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_1 = test.withColumn('prediction', lit(1.0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test set metrics:\n",
      "Accuracy: 0.20588235294117646\n",
      "F-1 Score:0.07030129124820661\n"
     ]
    }
   ],
   "source": [
    "evaluator = MulticlassClassificationEvaluator(predictionCol=\"prediction\")\n",
    "print('Test set metrics:')\n",
    "print('Accuracy: {}'.format(evaluator.evaluate(results_1, {evaluator.metricName: \"accuracy\"})))\n",
    "print('F-1 Score:{}'.format(evaluator.evaluate(results_1, {evaluator.metricName: \"f1\"})))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_0 = test.withColumn('prediction', lit(0.0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test set metrics:\n",
      "Accuracy: 0.7941176470588235\n",
      "F-1 Score:0.7029893924783028\n"
     ]
    }
   ],
   "source": [
    "evaluator = MulticlassClassificationEvaluator(predictionCol=\"prediction\")\n",
    "print('Test set metrics:')\n",
    "print('Accuracy: {}'.format(evaluator.evaluate(results_0, {evaluator.metricName: \"accuracy\"})))\n",
    "print('F-1 Score:{}'.format(evaluator.evaluate(results_0, {evaluator.metricName: \"f1\"})))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As a next step, we will perform classification with different algorithms to observe the result. \n",
    "\n",
    "[1] https://spark.apache.org/docs/latest/ml-classification-regression.html"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Gradient Boost and Random Forest"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "The idea of boosting came out of the idea of whether a weak learner can be modified to become better. A weak hypothesis or weak learner is defined as one whose performance is at least slightly better than random chance.\n",
    "Gradient Boost Trees is the modified and improved version of AdaBoost. \n",
    "\n",
    "AdaBoost, which is the abbreviation of the Adaptive Boosting, is the first practical boosting algorithm proposed by Freund and Schapire in 1996. AdaBoost is an algorithm for constructing a “strong” classifier as linear combination of “simple” “weak” classifier. In other words, the main working principle is to convert a set of weak classifiers into a strong one. Weak classifier is described as less than 50% error over any distribution and strong classifier is thresholded linear combination of the weak classifier outputs. AdaBoost works by weighting the observations, putting more weight on difficult to classify instances and less on those already handled well. New weak learners are added sequentially that focus their training on the more difficult patterns. Predictions are made by majority vote of the weak learners’ predictions, weighted by their individual accuracy. \n",
    "\n",
    "Gradient boosting involves three elements:\n",
    "* A loss function to be optimized. The loss function used depends on the type of problem being solved. It must be differentiable, but many standard loss functions are supported and you can define your own. For example, regression may use a squared error and classification may use logarithmic loss\n",
    "* A weak learner to make predictions. Decision trees are used as the weak learner in gradient boosting\n",
    "* An additive model to add weak learners to minimize the loss function. Trees are added one at a time, and existing trees in the model are not changed. A gradient descent procedure is used to minimize the loss when adding trees.\n",
    "\n",
    "GBT build trees one at a time, where each new tree helps to correct errors made by previously trained tree. A great application of GBM is anomaly detection in supervised learning settings where data is often highly unbalanced such as DNA sequences, credit card transactions or cybersecurity.\n",
    "\n",
    "The differences between the GBT and RF can be explained as Gradient boosting uses regression trees for prediction purpose where a random forest use decision tree. ... The random forest is easy to parallelize but boosted trees are hard to do. Random forests overfit a sample of the training data and then reduces the overfit by simple averaging the predictors. If you carefully tune parameters, gradient boosting can result in better performance than random forests. However, gradient boosting may not be a good choice if you have a lot of noise, as it can result in overfitting. They also tend to be harder to tune than random forests.\n",
    "\n",
    "[1] https://machinelearningmastery.com/gentle-introduction-gradient-boosting-algorithm-machine-learning/\n",
    "\n",
    "[2] https://medium.com/@aravanshad/gradient-boosting-versus-random-forest-cfa3fa8f0d80\n",
    "\n",
    "[3] https://www.datasciencecentral.com/profiles/blogs/decision-tree-vs-random-forest-vs-boosted-trees-explained"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "# initialize the GBT classifier\n",
    "gbt = GBTClassifier(maxIter = 10, seed = 45)\n",
    "\n",
    "# set evaluator\n",
    "f1_evaluator = MulticlassClassificationEvaluator(metricName='f1')\n",
    "\n",
    "# build paramGrid\n",
    "paramGrid = ParamGridBuilder().build()\n",
    "\n",
    "gbt_crossvalidation = CrossValidator(estimator = gbt,\n",
    "                                    estimatorParamMaps = paramGrid,\n",
    "                                    evaluator = f1_evaluator,\n",
    "                                    numFolds = 3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Fit the training data:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The training process took 1140.1361155509949 seconds\n"
     ]
    }
   ],
   "source": [
    "start = time()\n",
    "cvModel_gbt = gbt_crossvalidation.fit(train)\n",
    "end = time()\n",
    "cvModel_gbt.avgMetrics\n",
    "print('The training process took {} seconds'.format(end - start))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "gbt_result = cvModel_gbt.transform(validation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Gradient Boosted Trees Metrics:\n",
      "Accuracy: 0.8333333333333334\n",
      "F-1 Score:0.838095238095238\n"
     ]
    }
   ],
   "source": [
    "evaluator = MulticlassClassificationEvaluator(predictionCol = \"prediction\")\n",
    "print('Gradient Boosted Trees Metrics:')\n",
    "print('Accuracy: {}'.format(evaluator.evaluate(gbt_result, {evaluator.metricName: \"accuracy\"})))\n",
    "print('F-1 Score:{}'.format(evaluator.evaluate(gbt_result, {evaluator.metricName: \"f1\"})))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [],
   "source": [
    "# initialize the random forest classifier\n",
    "randomforest = RandomForestClassifier()\n",
    "\n",
    "# set evaluator\n",
    "f1_evaluator = MulticlassClassificationEvaluator(metricName = 'f1')\n",
    "\n",
    "# build paramGrid\n",
    "paramGrid = ParamGridBuilder().build()\n",
    "\n",
    "randomforest_crossval = CrossValidator(estimator = randomforest,\n",
    "                              estimatorParamMaps = paramGrid,\n",
    "                              evaluator = f1_evaluator,\n",
    "                              numFolds = 3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The training process took 520.41619181633 seconds\n"
     ]
    }
   ],
   "source": [
    "start = time()\n",
    "cvModel_rf = randomforest_crossval.fit(train)\n",
    "end = time()\n",
    "cvModel_rf.avgMetrics\n",
    "print('The training process took {} seconds'.format(end - start))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [],
   "source": [
    "randomforest_result = cvModel_rf.transform(validation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Random Forest Metrics:\n",
      "Accuracy: 1.0\n",
      "F-1 Score:1.0\n"
     ]
    }
   ],
   "source": [
    "evaluator = MulticlassClassificationEvaluator(predictionCol=\"prediction\")\n",
    "print('Random Forest Metrics:')\n",
    "print('Accuracy: {}'.format(evaluator.evaluate(randomforest_result, {evaluator.metricName: \"accuracy\"})))\n",
    "print('F-1 Score:{}'.format(evaluator.evaluate(randomforest_result, {evaluator.metricName: \"f1\"})))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Support Vector Machines\n",
    "The working principle of the support vector machines is to find or create the hyperplane in an N-dimensional space that distinctly classifies the data points. The dimension of the hyperplane is depending on the number of the features. Hyperplanes are basicly the desicion boundaries that helps the classification of the data such that data points which are locating at te different side of the hyperplane are classified as different classes. Support vector machines are commonly used in face detection, text and hypertext categorization, classification of images, bioinformatics and, handwriting recognition. One of the real world example is performed in the article \"Application of support vector machine modeling for prediction of common diseases: the case of diabetes and pre-diabetes\".\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [],
   "source": [
    "# initialize classifier\n",
    "svm = LinearSVC(maxIter = 10)\n",
    "\n",
    "# set evaluator\n",
    "f1_evaluator = MulticlassClassificationEvaluator(metricName='f1')\n",
    "\n",
    "# build paramGrid\n",
    "paramGrid = ParamGridBuilder().build()\n",
    "\n",
    "svm_crossval = CrossValidator(estimator=svm,\n",
    "                          estimatorParamMaps=paramGrid,\n",
    "                          evaluator=f1_evaluator,\n",
    "                          numFolds=3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The training process took 709.1884164810181 seconds\n"
     ]
    }
   ],
   "source": [
    "start = time()\n",
    "cvModel_svm = svm_crossval.fit(train)\n",
    "end = time()\n",
    "cvModel_svm.avgMetrics\n",
    "print('The training process took {} seconds'.format(end - start))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [],
   "source": [
    "svm_result = cvModel_svm.transform(validation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SVM Metrics:\n",
      "Accuracy: 0.6666666666666666\n",
      "F-1 Score:0.5333333333333333\n"
     ]
    }
   ],
   "source": [
    "evaluator = MulticlassClassificationEvaluator(predictionCol=\"prediction\")\n",
    "print('SVM Metrics:')\n",
    "print('Accuracy: {}'.format(evaluator.evaluate(svm_result, {evaluator.metricName: \"accuracy\"})))\n",
    "print('F-1 Score:{}'.format(evaluator.evaluate(svm_result, {evaluator.metricName: \"f1\"})))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Decision Tree Classifier\n",
    "\n",
    "In decision analysis, one of the visual and explicit representation of decision and decision making procedure can be performed by using decision trees. The decision tree is using a tree-like model of decisions such that a flowchart like tree structure in which each node internal node denotes a test and each branch represents an outcome of the test and each leaf node holds a corresponding class label. Decision tree is the one of the most powerfull and popular tool for classification and prediction. Decision trees are commonly using in medical diagnosis, failure prediction, credit scroing and crime risk investigation. Several business analysis examples can be found but one of the widely used one is the detection of the raudulent Financial Statements."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [],
   "source": [
    "dt = DecisionTreeClassifier()\n",
    "\n",
    "# set evaluator\n",
    "f1_evaluator = MulticlassClassificationEvaluator(metricName='f1')\n",
    "\n",
    "# build paramGrid\n",
    "paramGrid = ParamGridBuilder().build()\n",
    "\n",
    "dt_crossval = CrossValidator(estimator=dt,\n",
    "                          estimatorParamMaps=paramGrid,\n",
    "                          evaluator=f1_evaluator,\n",
    "                          numFolds=3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The training process took 595.8720190525055 seconds\n"
     ]
    }
   ],
   "source": [
    "start = time()\n",
    "cvModel_dt = dt_crossval.fit(train)\n",
    "end = time()\n",
    "cvModel_dt.avgMetrics\n",
    "print('The training process took {} seconds'.format(end - start))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [],
   "source": [
    "dt_result = cvModel_dt.transform(validation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SVM Metrics:\n",
      "Accuracy: 0.8333333333333334\n",
      "F-1 Score:0.838095238095238\n"
     ]
    }
   ],
   "source": [
    "evaluator = MulticlassClassificationEvaluator(predictionCol=\"prediction\")\n",
    "print('Decision Trees Metrics:')\n",
    "print('Accuracy: {}'.format(evaluator.evaluate(dt_result, {evaluator.metricName: \"accuracy\"})))\n",
    "print('F-1 Score:{}'.format(evaluator.evaluate(dt_result, {evaluator.metricName: \"f1\"})))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Model Comparision and Decision\n",
    "\n",
    "Gradient Boost Trees, Random Forest, Support Vector Machines, and Decision Tree algorithms are selected and used for training. Below is the summary of obtained accuracy, F1 Score during a recorded training time\n",
    "* **Gradient Boost Trees** gave the accuracy of **0.833**, F1 Score of **0.830** with **1140.13 secs** of training duration.\n",
    "\n",
    "* **Random Forest** gave the accuracy of **1.00**, F1 Score of **1.00** with **520.41 secs** of training duration.\n",
    "\n",
    "* **Support Vector Machines** gave the accuracy of **0.666**, F1 Score of **0.5333** with **709.8 secs** of training duration.\n",
    "\n",
    "* **Decision Trees** gave the accuracy of **0.833**, F1 Score of **0.830** with **595.87 secs** of training duration.\n",
    "\n",
    "Since the accuracy and F1 Score are the most important results for selecting the algorithm, training duration time has also an effect while working with large dataset. Since we have a relatively small dataset, I will choose the GBT algorithm because it has the 0.833% accuracy.\n",
    "\n",
    "Note: Random Forest algorithm provides 1.0 accuracy and F1 results which is interesting and can be sign of overfitting.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Model Tuning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [],
   "source": [
    "#gbt = GBTClassifier(maxIter = 10, seed = 45)\n",
    "\n",
    "# build paramGrid\n",
    "#paramGrid_gbt = ParamGridBuilder() \\\n",
    "#    .addGrid(gbt.maxDepth,[5, 10]) \\\n",
    "#    .addGrid(gbt.maxIter, [10,15]) \\\n",
    "#    .build()\n",
    "\n",
    "# set evaluator\n",
    "#f1_evaluator = MulticlassClassificationEvaluator(metricName='f1')\n",
    "\n",
    "#gbt_crossval = CrossValidator(estimator=gbt,\n",
    "#                          estimatorParamMaps=paramGrid_gbt,\n",
    "#                          evaluator=f1_evaluator,\n",
    "#                          numFolds=3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {},
   "outputs": [],
   "source": [
    "#start = time()\n",
    "#cvModel_gbt = gbt_crossval.fit(train)\n",
    "#end = time()\n",
    "#cvModel_gbt.avgMetrics\n",
    "#print('The training process took {} seconds'.format(end - start))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [],
   "source": [
    "#best_model = GBTClassifier(maxIter=10,seed=45, maxDepth=5)\n",
    "#best_model = best_model.fit(train)\n",
    "#results = best_model.transform(test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Model Metrics:\n",
      "Accuracy: 0.7647058823529411\n",
      "F-1 Score:0.7492997198879553\n"
     ]
    }
   ],
   "source": [
    "#evaluator = MulticlassClassificationEvaluator(predictionCol=\"prediction\")\n",
    "#print('Best Model Metrics:')\n",
    "#print('Accuracy: {}'.format(evaluator.evaluate(results, {evaluator.metricName: \"accuracy\"})))\n",
    "#print('F-1 Score:{}'.format(evaluator.evaluate(results, {evaluator.metricName: \"f1\"})))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {},
   "outputs": [],
   "source": [
    "# The Gradient-Boosted Trees (GBTs) model\n",
    "\n",
    "#model = GBTClassifier(seed = 45)\n",
    "#param_grid = ParamGridBuilder().addGrid(model.maxDepth,[3, 5, 10]).addGrid(model.maxBins ,[10, 5]) \\\n",
    "#    .addGrid(model.maxIter ,[20, 5]).build()\n",
    "\n",
    "#gbt_crossval = CrossValidator(estimator=model,\n",
    "#                          estimatorParamMaps=param_grid,\n",
    "#                          evaluator=f1_evaluator,\n",
    "#                          numFolds=3)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {},
   "outputs": [],
   "source": [
    "#start = time()\n",
    "#cvModel_gbt = gbt_crossval.fit(train)\n",
    "#end = time()\n",
    "#cvModel_dt.avgMetrics\n",
    "#print('The training process took {} seconds'.format(end - start))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#gbt_result = cvModel_gbt.transform(validation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#evaluator = MulticlassClassificationEvaluator(predictionCol=\"prediction\")\n",
    "#print('GBT Metrics:')\n",
    "#print('Accuracy: {}'.format(evaluator.evaluate(gbt_result, {evaluator.metricName: \"accuracy\"})))\n",
    "#print('F-1 Score:{}'.format(evaluator.evaluate(gbt_result, {evaluator.metricName: \"f1\"})))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Feature Importance Observation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {},
   "outputs": [],
   "source": [
    "gbt_best = GBTClassifier(maxIter=10,seed=45)\n",
    "gbt_best_model = gbt_best.fit(train)\n",
    "results_final = gbt_best_model.transform(test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test set metrics:\n",
      "Accuracy: 0.7647058823529411\n",
      "F-1 Score:0.7492997198879553\n"
     ]
    }
   ],
   "source": [
    "evaluator = MulticlassClassificationEvaluator(predictionCol=\"prediction\")\n",
    "print('Test set metrics:')\n",
    "print('Accuracy: {}'.format(evaluator.evaluate(results_final, {evaluator.metricName: \"accuracy\"})))\n",
    "print('F-1 Score:{}'.format(evaluator.evaluate(results_final, {evaluator.metricName: \"f1\"})))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f45081f0a90>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "features = gbt_best_model.featureImportances.values\n",
    "features_labels = ['gender', 'total_songs', 'thumbs_up', 'thumbs_down', 'add_to_playlist',\n",
    "                   'avg_songs_played', 'artist_count', 'add_friend', 'listen_time', 'lifetime', 'downgrade_label' ]\n",
    "y_pos = np.arange(len(features_labels))\n",
    "plt.figure(figsize=(10,8)) \n",
    "plt.barh(y_pos, features, align='center', color = 'magenta')\n",
    "plt.yticks(y_pos, features_labels)\n",
    "plt.xlabel('Importance Score')\n",
    "plt.title('Feature Importances');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Feature importance is utilized for Gradient Boost Tree model. From the above Figure, it is seen that thumbs down feature has a very important role. And second important feature is lifetime. It is reasonable that lifetime is an indicator because it is known that churned users have a shorter lifetime. \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Conclusion\n",
    "\n",
    "The machine learning modeling has a success in predicting the customer's churn activity. That can help the application owners to improve the user lifetime. Despite the relatively good results of other selected machine learning models such as SVM and Decision Trees, the GBT algorithm is selected and tuned with hyperparameters. For a future work, the Random Forest algorithm can be adjusted with different settings to overcome the over-fitting.\n",
    "\n",
    "###### Possible Improvements for Current Model\n",
    "Even if we have almost 80% accuracy in current model, there is still a chance to improve it by \n",
    "* Working with more observations by increasing the dataset size.\n",
    "* Investigating the different parameter settings to have more accuracy.\n",
    "* Analyzing location information effect on the churned user since location can provide us a information about life style and give company owner a chance to provide location based promotions to keep the users.\n",
    "* Used environment can also provide a usefull information. If android users are more likely to churned, this means the application needs an improvement for that environment.\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Final Steps\n",
    "Clean up your code, adding comments and renaming variables to make the code easier to read and maintain. Refer to the Spark Project Overview page and Data Scientist Capstone Project Rubric to make sure you are including all components of the capstone project and meet all expectations. Remember, this includes thorough documentation in a README file in a Github repository, as well as a web app or blog post."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}