Sample_EDA_Cleaning.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sample data Exploration Data Analysis & Further Cleaning\n",
    "### Since we randomly sampled from the original dataset, it's OK to do EDA in sampled datasets( from spark to pandas). This is done after built up first model\n",
    "\n",
    "### 1. sample_play\n",
    "Mainly for play_time/song_length"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import datetime as dt\n",
    "import matplotlib.pyplot as plt\n",
    "% matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "#1.Load in sample_play\n",
    "##notice the type of variables\n",
    "colname = ['uid', 'device', 'song_id', 'play_time', 'song_length', 'date', 'Churn']\n",
    "datatype = {'uid': 'int64', 'device': 'object', 'song_id': 'float64', 'song_length': 'float64', 'date':'object', 'Churn': 'int64'}\n",
    "sample_play = pd.read_csv(\"D:/MusicFile/sample_play_316.csv\", header = 0, dtype = datatype, names = colname)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "uid                     int64\n",
      "device                 object\n",
      "song_id               float64\n",
      "play_time             float64\n",
      "song_length           float64\n",
      "date           datetime64[ns]\n",
      "Churn                   int64\n",
      "dtype: object\n"
     ]
    },
    {
     "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>uid</th>\n",
       "      <th>device</th>\n",
       "      <th>song_id</th>\n",
       "      <th>play_time</th>\n",
       "      <th>song_length</th>\n",
       "      <th>date</th>\n",
       "      <th>Churn</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4879194</th>\n",
       "      <td>168519424</td>\n",
       "      <td>ar</td>\n",
       "      <td>1575088.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>275.0</td>\n",
       "      <td>2017-04-08</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>628184</th>\n",
       "      <td>168818752</td>\n",
       "      <td>ar</td>\n",
       "      <td>0.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>154.0</td>\n",
       "      <td>2017-04-04</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1502147</th>\n",
       "      <td>154595776</td>\n",
       "      <td>ar</td>\n",
       "      <td>3213921.0</td>\n",
       "      <td>263.0</td>\n",
       "      <td>263.0</td>\n",
       "      <td>2017-03-09</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4166495</th>\n",
       "      <td>167917008</td>\n",
       "      <td>ip</td>\n",
       "      <td>437869.0</td>\n",
       "      <td>229.0</td>\n",
       "      <td>229.0</td>\n",
       "      <td>2017-04-09</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3711529</th>\n",
       "      <td>168560560</td>\n",
       "      <td>ar</td>\n",
       "      <td>435280.0</td>\n",
       "      <td>183.0</td>\n",
       "      <td>284.0</td>\n",
       "      <td>2017-04-04</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4462629</th>\n",
       "      <td>167681200</td>\n",
       "      <td>ar</td>\n",
       "      <td>9823138.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>111.0</td>\n",
       "      <td>2017-04-02</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>750716</th>\n",
       "      <td>94891472</td>\n",
       "      <td>ar</td>\n",
       "      <td>626047.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>308.0</td>\n",
       "      <td>2017-03-30</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4177734</th>\n",
       "      <td>168924832</td>\n",
       "      <td>ip</td>\n",
       "      <td>6916106.0</td>\n",
       "      <td>78.0</td>\n",
       "      <td>236.0</td>\n",
       "      <td>2017-04-06</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               uid device    song_id  play_time  song_length       date  Churn\n",
       "4879194  168519424     ar  1575088.0        5.0        275.0 2017-04-08      0\n",
       "628184   168818752     ar        0.0       14.0        154.0 2017-04-04      0\n",
       "1502147  154595776     ar  3213921.0      263.0        263.0 2017-03-09      1\n",
       "4166495  167917008     ip   437869.0      229.0        229.0 2017-04-09      0\n",
       "3711529  168560560     ar   435280.0      183.0        284.0 2017-04-04      0\n",
       "4462629  167681200     ar  9823138.0        4.0        111.0 2017-04-02      0\n",
       "750716    94891472     ar   626047.0        0.0        308.0 2017-03-30      0\n",
       "4177734  168924832     ip  6916106.0       78.0        236.0 2017-04-06      0"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#check variable types\n",
    "sample_play['date'] = pd.to_datetime(sample_play['date'])\n",
    "print(sample_play.dtypes)\n",
    "\n",
    "sample_play.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "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>uid</th>\n",
       "      <th>song_id</th>\n",
       "      <th>date</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>49423096</td>\n",
       "      <td>4348548.0</td>\n",
       "      <td>2017-04-05</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>49423096</td>\n",
       "      <td>4348548.0</td>\n",
       "      <td>2017-04-05</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>77</th>\n",
       "      <td>49423096</td>\n",
       "      <td>4348548.0</td>\n",
       "      <td>2017-04-06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78</th>\n",
       "      <td>49423096</td>\n",
       "      <td>4348548.0</td>\n",
       "      <td>2017-04-06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>151</th>\n",
       "      <td>49423096</td>\n",
       "      <td>4348548.0</td>\n",
       "      <td>2017-04-07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4955453</th>\n",
       "      <td>168915904</td>\n",
       "      <td>6611804.0</td>\n",
       "      <td>2017-04-27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4955454</th>\n",
       "      <td>168915904</td>\n",
       "      <td>6611804.0</td>\n",
       "      <td>2017-04-27</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               uid    song_id       date\n",
       "24        49423096  4348548.0 2017-04-05\n",
       "25        49423096  4348548.0 2017-04-05\n",
       "77        49423096  4348548.0 2017-04-06\n",
       "78        49423096  4348548.0 2017-04-06\n",
       "151       49423096  4348548.0 2017-04-07\n",
       "4955453  168915904  6611804.0 2017-04-27\n",
       "4955454  168915904  6611804.0 2017-04-27"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#check whether each record records single play or cumulative play\n",
    "sample_play.groupby(['uid', 'song_id', 'date']).head()\n",
    "sample_play[['uid', 'song_id', 'date']][((sample_play.uid ==49423096)&(sample_play.song_id==4348548.0))|((sample_play.uid == 168915904)&(sample_play.song_id == 6611804.0))]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here, we can see that in same date, there are multiple records for same uid and same song_id, which enables me to make assumption as follows.\n",
    "**In spark, I only did roughly filter/impute of play_time and song_length, if just leave those data to feature generation, there might be difficult to create features about play_completition, so it's better to clean play_time and song_length... based on heuristic knowledge** \n",
    "\n",
    "My cleaning principle is: \n",
    "\n",
    "**Assumption**: each record only record one play time(not cumulative), if user play one song iteratively, the records will show multiple records, thus play_time should be ≤ song_length\n",
    "\n",
    "**play_time**:\n",
    "- if >10E5(only (20/4235452) no sense)(user = 17) or play_time == 0(user skipped play or clicked by error), drop them(user = 8)\n",
    "- if >10E4, devided by 1000. Since it's possible play_time was recorded as miliseconds\n",
    "- if > song_length, check its correctness, impute by the above way or song_length-1\n",
    "\n",
    "**song_length**:\n",
    "- if >10E6(no sense) or = 0, impute with play_time this step is done in Spark; by later check, find there is no song_length >10E4, which means existed song_length are reasonable, and imputed song_length has reasonable play_time. Play_time can be derived by song_length\n",
    "\n",
    "**[Notice]song_id**:\n",
    "- song_id ==0, with multiple song_length and play_time by different users, which means it's not a unique song, it's happened possibly due to mislocation of song_type(with value=0,1,2). There are nearly 10%(346692/4235452) of song_id =0. It's OK to just leave them as original, since we are now doing Churn prediction. However, if we want to do song_recommendation in the future, we should find better way to impute them. \n",
    "- song_id = -1(177/4235452), with 9 users. Also keep. \n",
    "\n",
    "#### 1.1 play_time & song_length"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "play_time >10E5: 0\n",
      "song_length>10E3 62\n",
      "check how many play_time > song_length and:\n",
      "play_time >10E5: 0\n",
      "play_time ∈(10E4, 10E5): 28147\n",
      "play_time ∈(10E3, 10E4): 7789\n",
      "play_time ∈(10E2, 10E3): 6850\n",
      "Check how many users have incorrect \n",
      "play_time >10E5: 0\n",
      "play_time ∈(10E4, 10E5): 15\n",
      "play_time ∈(10E3, 10E4): 15\n",
      "play_time ∈(10E2, 10E3): 18\n"
     ]
    }
   ],
   "source": [
    "print(\"play_time >10E5:\", sample_play.play_time[(sample_play.play_time>10E5)].count())\n",
    "print(\"song_length>10E3\", sample_play.play_time[(sample_play.song_length>10E3)].count())\n",
    "print(\"check how many play_time > song_length and:\")\n",
    "print(\"play_time >10E5:\", sample_play.play_time[(sample_play.play_time>10E5)&(sample_play.play_time>sample_play.song_length)].count())\n",
    "print(\"play_time ∈(10E4, 10E5):\", sample_play.song_length[((sample_play.play_time>10E4)&(sample_play.play_time<10E5))&(sample_play.play_time>sample_play.song_length)].count())\n",
    "print(\"play_time ∈(10E3, 10E4):\", sample_play.song_length[((sample_play.play_time>10E3)&(sample_play.play_time<10E4))&(sample_play.play_time>sample_play.song_length)].count())\n",
    "print(\"play_time ∈(10E2, 10E3):\", sample_play.song_length[((sample_play.play_time>10E2)&(sample_play.play_time<10E3))&(sample_play.play_time>sample_play.song_length)].count())\n",
    "print(\"Check how many users have incorrect \")\n",
    "print(\"play_time >10E5:\",sample_play.uid[sample_play.play_time[(sample_play.play_time>10E5)&(sample_play.play_time>sample_play.song_length)]].value_counts().shape[0])\n",
    "print(\"play_time ∈(10E4, 10E5):\",sample_play.uid[sample_play.song_length[((sample_play.play_time>10E4)&(sample_play.play_time<10E5))&(sample_play.play_time>sample_play.song_length)]].value_counts().shape[0])\n",
    "print(\"play_time ∈(10E3, 10E4):\",sample_play.uid[sample_play.song_length[((sample_play.play_time>10E3)&(sample_play.play_time<10E4))&(sample_play.play_time>sample_play.song_length)]].value_counts().shape[0])\n",
    "print(\"play_time ∈(10E2, 10E3):\",sample_play.uid[sample_play.song_length[((sample_play.play_time>10E2)&(sample_play.play_time<10E3))&(sample_play.play_time>sample_play.song_length)]].value_counts().shape[0])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "#filter sample_play by play_time and song_length\n",
    "#by play_time\n",
    "sample_play = sample_play[sample_play.play_time < 10E5]\n",
    "sample_play = sample_play[0<sample_play.play_time] \n",
    "#by song_Length"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#first convert song_length from float to int\n",
    "#sample_play.song_length = sample_play.song_length.astype(int)\n",
    "#len(str(sample_play.song_length[sample_play.iloc]))\n",
    "\n",
    "\n",
    "#sample_play['song_dig'] = sample_play['song_length'].apply(lambda x: len(str(x)))\n",
    "#sample_play['adjust_play_time'] = sample_play.apply(lambda x: x['play_time'][: x['song_dig']], axis = 1)\n",
    "###that's what I want to do, but couldn't because of Memory error!!!!\n",
    "#truncate play_time to len(song_length) if play_time >song_length\n",
    "#sample_play['adjust_play_time'] = sample_play.play_time.apply(lambda x: x[:])\n",
    "'''\n",
    "def trunc(df):\n",
    "    if df.play_time > df.song_length:\n",
    "        l = len(str(df.song_length))\n",
    "        df.play_time = df.play_time[:l]\n",
    "    else:\n",
    "        df.play_time = df.play_time\n",
    "    return df.play_time\n",
    "    '''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "#convert miliseconds => seconds\n",
    "sample_play['adjust_play_time'] = np.where(sample_play['play_time'] > 10E4, (sample_play['play_time']/1000).astype(int), sample_play['play_time'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "check how many play_time > song_length and:\n",
      "adjust_play_time >10E4: 0\n",
      "adjust_play_time ∈(10E2, 10E4): 14639\n",
      "Check how many users have incorrect \n",
      "adjust_play_time >10E4: 0\n",
      "adjust_play_time ∈(10E2, 10E4): 18\n",
      "adjust_play_time>song_length: 7059\n"
     ]
    }
   ],
   "source": [
    "#after milisecond=>seconds\n",
    "print(\"check how many play_time > song_length and:\")\n",
    "print(\"adjust_play_time >10E4:\", sample_play.play_time[(sample_play.adjust_play_time>10E4)&(sample_play.adjust_play_time>sample_play.song_length)].count())\n",
    "print(\"adjust_play_time ∈(10E2, 10E4):\", sample_play.song_length[((sample_play.adjust_play_time>10E2)&(sample_play.adjust_play_time<10E4))&(sample_play.adjust_play_time>sample_play.song_length)].count())\n",
    "\n",
    "print(\"Check how many users have incorrect \")\n",
    "print(\"adjust_play_time >10E4:\",sample_play.uid[sample_play.play_time[(sample_play.adjust_play_time>10E4)&(sample_play.adjust_play_time>sample_play.song_length)]].value_counts().shape[0])\n",
    "print(\"adjust_play_time ∈(10E2, 10E4):\",sample_play.uid[sample_play.song_length[((sample_play.adjust_play_time>10E2)&(sample_play.adjust_play_time<10E4))&(sample_play.adjust_play_time>sample_play.song_length)]].value_counts().shape[0])\n",
    "print(\"adjust_play_time>song_length:\",len(sample_play.uid[sample_play.adjust_play_time>sample_play.song_length].unique()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "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>uid</th>\n",
       "      <th>device</th>\n",
       "      <th>song_id</th>\n",
       "      <th>play_time</th>\n",
       "      <th>song_length</th>\n",
       "      <th>date</th>\n",
       "      <th>Churn</th>\n",
       "      <th>adjust_play_time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3403784</th>\n",
       "      <td>167803488</td>\n",
       "      <td>ar</td>\n",
       "      <td>856673.0</td>\n",
       "      <td>26030.0</td>\n",
       "      <td>114.0</td>\n",
       "      <td>2017-04-13</td>\n",
       "      <td>0</td>\n",
       "      <td>26030.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3559716</th>\n",
       "      <td>168073296</td>\n",
       "      <td>ar</td>\n",
       "      <td>294622.0</td>\n",
       "      <td>20434.0</td>\n",
       "      <td>351.0</td>\n",
       "      <td>2017-04-28</td>\n",
       "      <td>0</td>\n",
       "      <td>20434.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2537927</th>\n",
       "      <td>167583760</td>\n",
       "      <td>ar</td>\n",
       "      <td>23495222.0</td>\n",
       "      <td>36502.0</td>\n",
       "      <td>320.0</td>\n",
       "      <td>2017-04-04</td>\n",
       "      <td>0</td>\n",
       "      <td>36502.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1328407</th>\n",
       "      <td>168284048</td>\n",
       "      <td>ar</td>\n",
       "      <td>359861.0</td>\n",
       "      <td>53340.0</td>\n",
       "      <td>207.0</td>\n",
       "      <td>2017-04-01</td>\n",
       "      <td>0</td>\n",
       "      <td>53340.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1062710</th>\n",
       "      <td>167605664</td>\n",
       "      <td>ar</td>\n",
       "      <td>23529236.0</td>\n",
       "      <td>30669.0</td>\n",
       "      <td>139.0</td>\n",
       "      <td>2017-04-03</td>\n",
       "      <td>0</td>\n",
       "      <td>30669.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4762532</th>\n",
       "      <td>167945200</td>\n",
       "      <td>ar</td>\n",
       "      <td>708072.0</td>\n",
       "      <td>86378.0</td>\n",
       "      <td>326.0</td>\n",
       "      <td>2017-04-02</td>\n",
       "      <td>0</td>\n",
       "      <td>86378.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3515419</th>\n",
       "      <td>168685696</td>\n",
       "      <td>ar</td>\n",
       "      <td>158864.0</td>\n",
       "      <td>93015.0</td>\n",
       "      <td>276.0</td>\n",
       "      <td>2017-04-04</td>\n",
       "      <td>0</td>\n",
       "      <td>93015.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1458018</th>\n",
       "      <td>159405616</td>\n",
       "      <td>ar</td>\n",
       "      <td>23037386.0</td>\n",
       "      <td>59257.0</td>\n",
       "      <td>251.0</td>\n",
       "      <td>2017-04-09</td>\n",
       "      <td>0</td>\n",
       "      <td>59257.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3671391</th>\n",
       "      <td>154600016</td>\n",
       "      <td>ar</td>\n",
       "      <td>6358424.0</td>\n",
       "      <td>17335.0</td>\n",
       "      <td>241.0</td>\n",
       "      <td>2017-03-01</td>\n",
       "      <td>1</td>\n",
       "      <td>17335.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1162953</th>\n",
       "      <td>168643024</td>\n",
       "      <td>ar</td>\n",
       "      <td>858008.0</td>\n",
       "      <td>17987.0</td>\n",
       "      <td>93.0</td>\n",
       "      <td>2017-04-18</td>\n",
       "      <td>1</td>\n",
       "      <td>17987.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               uid device     song_id  play_time  song_length       date  \\\n",
       "3403784  167803488     ar    856673.0    26030.0        114.0 2017-04-13   \n",
       "3559716  168073296     ar    294622.0    20434.0        351.0 2017-04-28   \n",
       "2537927  167583760     ar  23495222.0    36502.0        320.0 2017-04-04   \n",
       "1328407  168284048     ar    359861.0    53340.0        207.0 2017-04-01   \n",
       "1062710  167605664     ar  23529236.0    30669.0        139.0 2017-04-03   \n",
       "4762532  167945200     ar    708072.0    86378.0        326.0 2017-04-02   \n",
       "3515419  168685696     ar    158864.0    93015.0        276.0 2017-04-04   \n",
       "1458018  159405616     ar  23037386.0    59257.0        251.0 2017-04-09   \n",
       "3671391  154600016     ar   6358424.0    17335.0        241.0 2017-03-01   \n",
       "1162953  168643024     ar    858008.0    17987.0         93.0 2017-04-18   \n",
       "\n",
       "         Churn  adjust_play_time  \n",
       "3403784      0           26030.0  \n",
       "3559716      0           20434.0  \n",
       "2537927      0           36502.0  \n",
       "1328407      0           53340.0  \n",
       "1062710      0           30669.0  \n",
       "4762532      0           86378.0  \n",
       "3515419      0           93015.0  \n",
       "1458018      0           59257.0  \n",
       "3671391      1           17335.0  \n",
       "1162953      1           17987.0  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_play[(sample_play.adjust_play_time>sample_play.song_length)&((sample_play.adjust_play_time>10E3)&(sample_play.adjust_play_time<10E5))].sample(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "After convert, Only 18 users have adjust play time ∈(10E2, 10E4), but 7059 users have adjust_play_time>song_length. By checking, find there are situations that not make too much sense...but we do notcie there are play_time: 58614 with song_length 59, play_time:99938 with song_length 100/ play_time:12075 with song_length 245/play_time:73933 with song_length 123. For the last two situation, It's kind of difficult to impute without further information. \n",
    "\n",
    "Thus, in order to impute all these three conditions, impute incorrect play_time with song_length -1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>uid</th>\n",
       "      <th>device</th>\n",
       "      <th>song_id</th>\n",
       "      <th>play_time</th>\n",
       "      <th>song_length</th>\n",
       "      <th>date</th>\n",
       "      <th>Churn</th>\n",
       "      <th>adjust_play_time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2884338</th>\n",
       "      <td>168571440</td>\n",
       "      <td>ar</td>\n",
       "      <td>7010204.0</td>\n",
       "      <td>66.0</td>\n",
       "      <td>65.0</td>\n",
       "      <td>2017-03-31</td>\n",
       "      <td>0</td>\n",
       "      <td>64.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4550605</th>\n",
       "      <td>154438112</td>\n",
       "      <td>ar</td>\n",
       "      <td>15195572.0</td>\n",
       "      <td>255.0</td>\n",
       "      <td>125.0</td>\n",
       "      <td>2017-03-06</td>\n",
       "      <td>1</td>\n",
       "      <td>124.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1066751</th>\n",
       "      <td>168364880</td>\n",
       "      <td>ar</td>\n",
       "      <td>4859593.0</td>\n",
       "      <td>267.0</td>\n",
       "      <td>266.0</td>\n",
       "      <td>2017-04-09</td>\n",
       "      <td>0</td>\n",
       "      <td>265.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3356699</th>\n",
       "      <td>167616208</td>\n",
       "      <td>ar</td>\n",
       "      <td>23498554.0</td>\n",
       "      <td>229878.0</td>\n",
       "      <td>230.0</td>\n",
       "      <td>2017-03-31</td>\n",
       "      <td>1</td>\n",
       "      <td>229.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1029956</th>\n",
       "      <td>154586096</td>\n",
       "      <td>ar</td>\n",
       "      <td>985258.0</td>\n",
       "      <td>194.0</td>\n",
       "      <td>193.0</td>\n",
       "      <td>2017-03-02</td>\n",
       "      <td>1</td>\n",
       "      <td>192.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>750908</th>\n",
       "      <td>94891472</td>\n",
       "      <td>ar</td>\n",
       "      <td>125223.0</td>\n",
       "      <td>255.0</td>\n",
       "      <td>254.0</td>\n",
       "      <td>2017-04-02</td>\n",
       "      <td>0</td>\n",
       "      <td>253.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>707005</th>\n",
       "      <td>82485712</td>\n",
       "      <td>ar</td>\n",
       "      <td>23375964.0</td>\n",
       "      <td>184.0</td>\n",
       "      <td>183.0</td>\n",
       "      <td>2017-04-18</td>\n",
       "      <td>0</td>\n",
       "      <td>182.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1834626</th>\n",
       "      <td>168789792</td>\n",
       "      <td>ar</td>\n",
       "      <td>0.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2017-03-31</td>\n",
       "      <td>1</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               uid device     song_id  play_time  song_length       date  \\\n",
       "2884338  168571440     ar   7010204.0       66.0         65.0 2017-03-31   \n",
       "4550605  154438112     ar  15195572.0      255.0        125.0 2017-03-06   \n",
       "1066751  168364880     ar   4859593.0      267.0        266.0 2017-04-09   \n",
       "3356699  167616208     ar  23498554.0   229878.0        230.0 2017-03-31   \n",
       "1029956  154586096     ar    985258.0      194.0        193.0 2017-03-02   \n",
       "750908    94891472     ar    125223.0      255.0        254.0 2017-04-02   \n",
       "707005    82485712     ar  23375964.0      184.0        183.0 2017-04-18   \n",
       "1834626  168789792     ar         0.0        6.0          4.0 2017-03-31   \n",
       "\n",
       "         Churn  adjust_play_time  \n",
       "2884338      0              64.0  \n",
       "4550605      1             124.0  \n",
       "1066751      0             265.0  \n",
       "3356699      1             229.0  \n",
       "1029956      1             192.0  \n",
       "750908       0             253.0  \n",
       "707005       0             182.0  \n",
       "1834626      1               3.0  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_play['adjust_play_time'] = np.where(sample_play.adjust_play_time>sample_play.song_length,sample_play.song_length-1, sample_play.adjust_play_time)\n",
    "sample_play[(sample_play.play_time>sample_play.song_length)].sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "#do not save row index\n",
    "sample_play.to_csv(\"D:/MusicFile/sample_play_318.csv\", header = True, index = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "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>adjust_play_time</th>\n",
       "      <th>song_length</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>4.235432e+06</td>\n",
       "      <td>4.235432e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>1.734569e+02</td>\n",
       "      <td>2.635448e+02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>1.849254e+02</td>\n",
       "      <td>2.850556e+02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>4.500000e+01</td>\n",
       "      <td>2.060000e+02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>1.930000e+02</td>\n",
       "      <td>2.430000e+02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>2.520000e+02</td>\n",
       "      <td>2.800000e+02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>4.189700e+04</td>\n",
       "      <td>6.685100e+04</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       adjust_play_time   song_length\n",
       "count      4.235432e+06  4.235432e+06\n",
       "mean       1.734569e+02  2.635448e+02\n",
       "std        1.849254e+02  2.850556e+02\n",
       "min        0.000000e+00  1.000000e+00\n",
       "25%        4.500000e+01  2.060000e+02\n",
       "50%        1.930000e+02  2.430000e+02\n",
       "75%        2.520000e+02  2.800000e+02\n",
       "max        4.189700e+04  6.685100e+04"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_play[['adjust_play_time', 'song_length']].describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1.2 Plot sample_play"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\users\\tianyi fang\\appdata\\local\\programs\\python\\python36-32\\lib\\site-packages\\numpy\\lib\\arraysetops.py:472: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  mask |= (ar1 == a)\n"
     ]
    }
   ],
   "source": [
    "datatype = {'uid': 'int64', 'device': 'object', 'song_id': 'float64', 'song_length': 'float64', 'date':'object', 'Churn': 'int64', 'adjust_play_time':'float64'}\n",
    "#colname = []\n",
    "sample_play = pd.read_csv(\"D:/MusicFile/sample_play_318.csv\", header = 0, dtype = datatype, index_col = 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "song_id                    float64\n",
      "song_length                float64\n",
      "date                datetime64[ns]\n",
      "Churn                        int64\n",
      "adjust_play_time           float64\n",
      "device                      object\n",
      "dtype: object\n"
     ]
    },
    {
     "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>song_id</th>\n",
       "      <th>song_length</th>\n",
       "      <th>date</th>\n",
       "      <th>Churn</th>\n",
       "      <th>adjust_play_time</th>\n",
       "      <th>device</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>uid</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>169022784</th>\n",
       "      <td>3345090.0</td>\n",
       "      <td>240.0</td>\n",
       "      <td>2017-04-02</td>\n",
       "      <td>0</td>\n",
       "      <td>239.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168126384</th>\n",
       "      <td>708770.0</td>\n",
       "      <td>265.0</td>\n",
       "      <td>2017-04-15</td>\n",
       "      <td>0</td>\n",
       "      <td>265.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154785136</th>\n",
       "      <td>125802.0</td>\n",
       "      <td>241.0</td>\n",
       "      <td>2017-03-05</td>\n",
       "      <td>1</td>\n",
       "      <td>9.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154466784</th>\n",
       "      <td>20273752.0</td>\n",
       "      <td>225.0</td>\n",
       "      <td>2017-03-06</td>\n",
       "      <td>1</td>\n",
       "      <td>224.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168518208</th>\n",
       "      <td>20674552.0</td>\n",
       "      <td>286.0</td>\n",
       "      <td>2017-04-21</td>\n",
       "      <td>0</td>\n",
       "      <td>286.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167646512</th>\n",
       "      <td>82445.0</td>\n",
       "      <td>271.0</td>\n",
       "      <td>2017-04-04</td>\n",
       "      <td>0</td>\n",
       "      <td>270.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168611904</th>\n",
       "      <td>505795.0</td>\n",
       "      <td>286.0</td>\n",
       "      <td>2017-04-04</td>\n",
       "      <td>0</td>\n",
       "      <td>286.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168246336</th>\n",
       "      <td>1566133.0</td>\n",
       "      <td>288.0</td>\n",
       "      <td>2017-04-12</td>\n",
       "      <td>0</td>\n",
       "      <td>288.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              song_id  song_length       date  Churn  adjust_play_time device\n",
       "uid                                                                          \n",
       "169022784   3345090.0        240.0 2017-04-02      0             239.0   both\n",
       "168126384    708770.0        265.0 2017-04-15      0             265.0     ar\n",
       "154785136    125802.0        241.0 2017-03-05      1               9.0   both\n",
       "154466784  20273752.0        225.0 2017-03-06      1             224.0     ar\n",
       "168518208  20674552.0        286.0 2017-04-21      0             286.0   both\n",
       "167646512     82445.0        271.0 2017-04-04      0             270.0     ar\n",
       "168611904    505795.0        286.0 2017-04-04      0             286.0   both\n",
       "168246336   1566133.0        288.0 2017-04-12      0             288.0     ar"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_play['date'] = pd.to_datetime(sample_play['date'])\n",
    "print(sample_play.dtypes)\n",
    "sample_play.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x22353690>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xe6ddf10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "q97 = sample_play[['adjust_play_time', 'song_length']][sample_play['song_length']<sample_play['song_length'].quantile(0.97)]\n",
    "q97 = pd.DataFrame(q97)\n",
    "fig, ax = plt.subplots()\n",
    "\n",
    "ax.hist(q97['adjust_play_time'], bins =30, alpha = 0.9, color = 'lightblue')\n",
    "ax.hist(q97['song_length'], bins =30, alpha = 0.5, color = 'salmon')\n",
    "plt.axvline(q97['adjust_play_time'].mean(), color='lightblue', linestyle='dashed', linewidth=2)\n",
    "plt.axvline(q97['song_length'].mean(), color='salmon', linestyle='dashed', linewidth=2)\n",
    "plt.title('Histogram of 97% song_length and adjust_play_time' , fontsize = 13)\n",
    "plt.xlabel(\"seconds\", fontsize = 13)\n",
    "plt.legend(loc='upper right', fontsize = 13)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "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>uid</th>\n",
       "      <th>song_id</th>\n",
       "      <th>song_length</th>\n",
       "      <th>date</th>\n",
       "      <th>Churn</th>\n",
       "      <th>adjust_play_time</th>\n",
       "      <th>device</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>49423096</td>\n",
       "      <td>203139.0</td>\n",
       "      <td>270.0</td>\n",
       "      <td>2017-04-02</td>\n",
       "      <td>0</td>\n",
       "      <td>204.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>49423096</td>\n",
       "      <td>133946.0</td>\n",
       "      <td>286.0</td>\n",
       "      <td>2017-04-02</td>\n",
       "      <td>0</td>\n",
       "      <td>165.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>49423096</td>\n",
       "      <td>380389.0</td>\n",
       "      <td>233.0</td>\n",
       "      <td>2017-04-02</td>\n",
       "      <td>0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>49423096</td>\n",
       "      <td>23534240.0</td>\n",
       "      <td>378.0</td>\n",
       "      <td>2017-04-02</td>\n",
       "      <td>0</td>\n",
       "      <td>115.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>49423096</td>\n",
       "      <td>6672344.0</td>\n",
       "      <td>249.0</td>\n",
       "      <td>2017-04-02</td>\n",
       "      <td>0</td>\n",
       "      <td>69.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        uid     song_id  song_length       date  Churn  adjust_play_time  \\\n",
       "0  49423096    203139.0        270.0 2017-04-02      0             204.0   \n",
       "1  49423096    133946.0        286.0 2017-04-02      0             165.0   \n",
       "2  49423096    380389.0        233.0 2017-04-02      0              40.0   \n",
       "3  49423096  23534240.0        378.0 2017-04-02      0             115.0   \n",
       "4  49423096   6672344.0        249.0 2017-04-02      0              69.0   \n",
       "\n",
       "  device  \n",
       "0     ar  \n",
       "1     ar  \n",
       "2     ar  \n",
       "3     ar  \n",
       "4     ar  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_play_d = sample_play.copy()\n",
    "sample_play_d.reset_index(inplace = True)\n",
    "sample_play_d.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "#check device\n",
    "device_count = sample_play_d[['uid', 'device']].drop_duplicates() \\\n",
    "                                              .groupby('uid')['device'] \\\n",
    "                                              .agg(['max', 'count'])\n",
    "device_count['device'] = np.where(device_count['count'] == 1, device_count['max'], 'both')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2ba7f90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "fig = plt.figure()\n",
    "ax = fig.add_subplot(111)\n",
    "dc = device_count['device'].value_counts()\n",
    "dcsum = dc.sum()\n",
    "dc.plot(kind = 'bar', ax= ax)\n",
    "plt.title('Distribution of each device type' , fontsize = 14)\n",
    "plt.xlabel(\"device\", fontsize = 14)\n",
    "plt.ylabel(\"Count\", fontsize = 14)\n",
    "plt.xticks( fontsize = 14)\n",
    "plt.yticks(fontsize = 14)\n",
    "plt.legend(loc='upper right', fontsize = 14)\n",
    "for i,v in dc.reset_index().iterrows():\n",
    "    ax.text(i,v['device']+1, v['device'],ha='left', va='bottom', fontsize = 12)\n",
    "    ax.text(i,(v['device']/dcsum).round(3) + 1, (v['device']/dcsum).round(3),ha='center',fontsize = 12)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "            adjust_play_time  song_length\n",
      "date                                     \n",
      "2017-03-01        127.220898   259.168609\n",
      "2017-03-02        164.391645   263.297592\n",
      "2017-03-03        171.543604   265.285909\n",
      "2017-03-04        174.524416   267.063617\n",
      "2017-03-05        178.442732   265.346159\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x3061e50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#play_time trend\n",
    "fig = plt.figure(figsize = (12,6))\n",
    "ax = fig.add_subplot(111)\n",
    "gb = sample_play.groupby(['date'])[['adjust_play_time','song_length']].mean()\n",
    "#gb['count']= sample_play['date'].value_counts()\n",
    "gb.plot(kind = 'line', ax = ax)\n",
    "plt.title('Trend of play_time and song_length', fontsize = 18)\n",
    "plt.xlabel('Date', fontsize = 15)\n",
    "plt.xticks(rotation = 'vertical')\n",
    "plt.ylabel('count', fontsize = 15)\n",
    "plt.xticks(fontsize = 14)\n",
    "plt.yticks(fontsize = 14)\n",
    "plt.legend(loc='center left', bbox_to_anchor=(1, 0.5), fontsize = 13)\n",
    "print(gb.head())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "index      int64\n",
      "uid        int64\n",
      "max       object\n",
      "count      int64\n",
      "device    object\n",
      "dtype: object\n"
     ]
    },
    {
     "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>index</th>\n",
       "      <th>uid</th>\n",
       "      <th>max</th>\n",
       "      <th>count</th>\n",
       "      <th>device</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>393</th>\n",
       "      <td>393</td>\n",
       "      <td>109958960</td>\n",
       "      <td>ar</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>427</th>\n",
       "      <td>427</td>\n",
       "      <td>118366104</td>\n",
       "      <td>ar</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4859</th>\n",
       "      <td>4859</td>\n",
       "      <td>167856400</td>\n",
       "      <td>ar</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8743</th>\n",
       "      <td>8743</td>\n",
       "      <td>169052448</td>\n",
       "      <td>ip</td>\n",
       "      <td>1</td>\n",
       "      <td>ip</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>958</th>\n",
       "      <td>958</td>\n",
       "      <td>154413392</td>\n",
       "      <td>ip</td>\n",
       "      <td>2</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3959</th>\n",
       "      <td>3959</td>\n",
       "      <td>167599616</td>\n",
       "      <td>ar</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>98</th>\n",
       "      <td>98</td>\n",
       "      <td>28055436</td>\n",
       "      <td>ar</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3198</th>\n",
       "      <td>3198</td>\n",
       "      <td>155084816</td>\n",
       "      <td>ip</td>\n",
       "      <td>1</td>\n",
       "      <td>ip</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      index        uid max  count device\n",
       "393     393  109958960  ar      1     ar\n",
       "427     427  118366104  ar      1     ar\n",
       "4859   4859  167856400  ar      1     ar\n",
       "8743   8743  169052448  ip      1     ip\n",
       "958     958  154413392  ip      2   both\n",
       "3959   3959  167599616  ar      1     ar\n",
       "98       98   28055436  ar      1     ar\n",
       "3198   3198  155084816  ip      1     ip"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "device_count.reset_index(inplace = True)\n",
    "print(device_count.dtypes)\n",
    "device_count.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>uid</th>\n",
       "      <th>device</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1595</th>\n",
       "      <td>154532160</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3898</th>\n",
       "      <td>167583216</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6621</th>\n",
       "      <td>168399952</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2765</th>\n",
       "      <td>154766544</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8733</th>\n",
       "      <td>169041984</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8231</th>\n",
       "      <td>168887056</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2898</th>\n",
       "      <td>154791456</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3960</th>\n",
       "      <td>167599792</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            uid device\n",
       "1595  154532160   both\n",
       "3898  167583216     ar\n",
       "6621  168399952     ar\n",
       "2765  154766544   both\n",
       "8733  169041984   both\n",
       "8231  168887056   both\n",
       "2898  154791456     ar\n",
       "3960  167599792     ar"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "device_df = device_count.drop('max',1).drop('count',1).drop('index',1)\n",
    "device_df.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "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>uid</th>\n",
       "      <th>song_id</th>\n",
       "      <th>play_time</th>\n",
       "      <th>song_length</th>\n",
       "      <th>date</th>\n",
       "      <th>Churn</th>\n",
       "      <th>adjust_play_time</th>\n",
       "      <th>device</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4127917</th>\n",
       "      <td>168288096</td>\n",
       "      <td>234717.0</td>\n",
       "      <td>240.0</td>\n",
       "      <td>267.0</td>\n",
       "      <td>2017-04-14</td>\n",
       "      <td>0</td>\n",
       "      <td>240.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1438091</th>\n",
       "      <td>168599568</td>\n",
       "      <td>6506486.0</td>\n",
       "      <td>264.0</td>\n",
       "      <td>264.0</td>\n",
       "      <td>2017-04-07</td>\n",
       "      <td>0</td>\n",
       "      <td>264.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>405289</th>\n",
       "      <td>167887840</td>\n",
       "      <td>5885319.0</td>\n",
       "      <td>307.0</td>\n",
       "      <td>308.0</td>\n",
       "      <td>2017-04-18</td>\n",
       "      <td>0</td>\n",
       "      <td>307.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3949566</th>\n",
       "      <td>168353312</td>\n",
       "      <td>5835282.0</td>\n",
       "      <td>411.0</td>\n",
       "      <td>410.0</td>\n",
       "      <td>2017-04-11</td>\n",
       "      <td>0</td>\n",
       "      <td>409.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2860225</th>\n",
       "      <td>168367200</td>\n",
       "      <td>55219.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>297.0</td>\n",
       "      <td>2017-04-07</td>\n",
       "      <td>0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>204283</th>\n",
       "      <td>168199440</td>\n",
       "      <td>130594.0</td>\n",
       "      <td>268.0</td>\n",
       "      <td>268.0</td>\n",
       "      <td>2017-04-09</td>\n",
       "      <td>0</td>\n",
       "      <td>268.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2036125</th>\n",
       "      <td>168618464</td>\n",
       "      <td>688056.0</td>\n",
       "      <td>244.0</td>\n",
       "      <td>245.0</td>\n",
       "      <td>2017-04-07</td>\n",
       "      <td>0</td>\n",
       "      <td>244.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2684201</th>\n",
       "      <td>168531536</td>\n",
       "      <td>914649.0</td>\n",
       "      <td>283.0</td>\n",
       "      <td>283.0</td>\n",
       "      <td>2017-03-31</td>\n",
       "      <td>0</td>\n",
       "      <td>283.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               uid    song_id  play_time  song_length       date  Churn  \\\n",
       "4127917  168288096   234717.0      240.0        267.0 2017-04-14      0   \n",
       "1438091  168599568  6506486.0      264.0        264.0 2017-04-07      0   \n",
       "405289   167887840  5885319.0      307.0        308.0 2017-04-18      0   \n",
       "3949566  168353312  5835282.0      411.0        410.0 2017-04-11      0   \n",
       "2860225  168367200    55219.0        6.0        297.0 2017-04-07      0   \n",
       "204283   168199440   130594.0      268.0        268.0 2017-04-09      0   \n",
       "2036125  168618464   688056.0      244.0        245.0 2017-04-07      0   \n",
       "2684201  168531536   914649.0      283.0        283.0 2017-03-31      0   \n",
       "\n",
       "         adjust_play_time device  \n",
       "4127917             240.0   both  \n",
       "1438091             264.0   both  \n",
       "405289              307.0     ar  \n",
       "3949566             409.0   both  \n",
       "2860225               6.0   both  \n",
       "204283              268.0   both  \n",
       "2036125             244.0   both  \n",
       "2684201             283.0   both  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#sample_play_d = sample_play.merge(device_df, on = 'uid', how = 'left')\n",
    "sample_play_d = sample_play_d.drop('device_x',1).rename(columns = {'device_y':'device'})\n",
    "#sample_play_d = sample_play_d.rename(columns = {'device_y':'device'})\n",
    "sample_play_d.sample(8)\n",
    "#sample_play.groupby(['date','device'])[['adjust_play_time','song_length']].mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4235432, 8)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_play_d.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "#do not save row index\n",
    "sample_play_d = sample_play_d.drop('play_time', 1)\n",
    "sample_play_d.to_csv(\"D:/MusicFile/sample_play_318.csv\", header = True, index = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['uid', 'song_id', 'song_length', 'date', 'Churn', 'adjust_play_time',\n",
       "       'device'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_play_d.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "device              ar        both          ip\n",
      "date                                          \n",
      "2017-03-01  130.878785  125.045935   72.373057\n",
      "2017-03-02  166.203492  164.563062   77.137253\n",
      "2017-03-03  176.045742  170.447218   88.071762\n",
      "2017-03-04  185.640174  168.831684  122.649772\n",
      "2017-03-05  184.642318  174.529049  170.789272\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x317a790>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#play_time trend of different devices\n",
    "fig = plt.figure(figsize = (12,6))\n",
    "ax = fig.add_subplot(111)\n",
    "gb = sample_play_d.groupby(['date','device'])['adjust_play_time'].mean().unstack()\n",
    "#gb['count']= sample_play['date'].value_counts()\n",
    "gb.plot(kind = 'line', ax = ax)\n",
    "plt.title('Trend of average play_time for different devices', fontsize = 18)\n",
    "plt.xlabel('Date', fontsize = 15)\n",
    "plt.xticks(rotation = 'vertical')\n",
    "plt.ylabel('count', fontsize = 15)\n",
    "plt.xticks(fontsize = 14)\n",
    "plt.yticks(fontsize = 14)\n",
    "plt.legend(loc='center left', bbox_to_anchor=(1, 0.5), fontsize = 13)\n",
    "print(gb.head())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As we can see, even we try to make play_time more sensible by boldly deviding it by either 1000 or 100, but there are still 6660 records(899 users) with adjust_play_time > song_length..It's really difficult to do further imputation without any further information, so I would like to just leave those records.\n",
    "\n",
    "**Notice** there is song_id ==0(346692/4235452), but with different song_length and different play_time of differnt user, which indicates song_id ==0 is not a unique song, there might be something wrong during record, since the next column is song_type(with value 0,1,2). check if drop song_id ==0, how many uid will be filtered.\n",
    "\n",
    "#### 1.2 song_id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "song_id = 0 346692\n",
      "song_id = -1 177\n"
     ]
    }
   ],
   "source": [
    "print(\"song_id = 0\",sample_play.uid[sample_play.song_id ==0].count())\n",
    "print(\"song_id = -1\",sample_play.uid[sample_play.song_id ==-1].count())\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compare with uncleaned:\n",
    "original rows: 4955455, after cleaning 4235432\n",
    "\n",
    "original churn rate:4465/4358, after 4464/4334"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8798, 2)"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_uid_label_8798 = sample_play_d[['uid','Churn']].drop_duplicates()\n",
    "sample_uid_label_8798.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "sample_uid_label_8798.to_csv(\"D:/MusicFile/sample_uid_label8798.csv\", header = True, index = False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## END"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}