Feature_Generation.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Feature Generation \n",
    "## 1.for sampled_play\n",
    "**UPDATE 3.18** sample_play_318 contains device(ip,ar, both) with adjust_play_time\n",
    "### 1.1 Frequency of time_period, average play_time of the same time_period, etc "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import datetime as dt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "% matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Load in sample_play\n",
    "**[Before 3.16]**Get the sample_uid_label(8823) by merge sample_uid(8823) with sample_uid_label(8863)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "uid    8823\n",
      "dtype: int64 uid      8862\n",
      "churn    8862\n",
      "dtype: int64\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>churn</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>177</th>\n",
       "      <td>168111040.0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>662</th>\n",
       "      <td>167676176.0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1924</th>\n",
       "      <td>154560160.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2320</th>\n",
       "      <td>167988528.0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85</th>\n",
       "      <td>168615296.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1081</th>\n",
       "      <td>168696800.0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6535</th>\n",
       "      <td>154525376.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5503</th>\n",
       "      <td>154741152.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              uid  churn\n",
       "177   168111040.0      0\n",
       "662   167676176.0      0\n",
       "1924  154560160.0      1\n",
       "2320  167988528.0      0\n",
       "85    168615296.0      1\n",
       "1081  168696800.0      0\n",
       "6535  154525376.0      1\n",
       "5503  154741152.0      1"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "s8823 = pd.read_csv(\"D:/MusicFile/sample_uid_316/sample_uid_8823.csv\")\n",
    "sl8862 = pd.read_csv(\"D:/MusicFile/sample_uid_label/sample_uid_316.csv\")\n",
    "sl8862.sample(8)\n",
    "print(s8823.count(), sl8862.count())\n",
    "sample_uid_label = s8823.merge(sl8862, on = 'uid', how = 'inner')\n",
    "\n",
    "sample_uid_label.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {},
   "outputs": [],
   "source": [
    "sample_uid_label.to_csv(\"D:/MusicFile/sample_uid_label8823.csv\", header = True, index = False)"
   ]
  },
  {
   "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>churn</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>uid</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>168706400.0</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18671898.0</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154413408.0</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168321472.0</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154817520.0</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168083408.0</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168574048.0</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167842240.0</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             churn\n",
       "uid               \n",
       "168706400.0      0\n",
       "18671898.0       1\n",
       "154413408.0      1\n",
       "168321472.0      0\n",
       "154817520.0      1\n",
       "168083408.0      0\n",
       "168574048.0      0\n",
       "167842240.0      0"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_uid_label=pd.read_csv(\"D:/MusicFile/sample_uid_label8823.csv\", header = 0, index_col = 0)\n",
    "sample_uid_label.sample(8)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load in 318 sample_play, uid_label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "#notice the type of variables\n",
    "colname = ['uid', 'song_id', 'song_length', 'date', 'Churn', 'play_time',\n",
    "       'device']\n",
    "datatype = {'uid': 'int64', 'Churn': 'int64'}\n",
    "sample_play = pd.read_csv(\"D:/MusicFile/sample_play_318.csv\", header = 0, dtype = datatype, names = colname)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8798, 1)"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_uid_8798 = pd.read_csv(\"D:/MusicFile/sample_uid_label8798.csv\", header = 0, index_col = 0)\n",
    "sample_uid_8798.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>play_time</th>\n",
       "      <th>device</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3179182</th>\n",
       "      <td>168559712</td>\n",
       "      <td>611618.0</td>\n",
       "      <td>235.0</td>\n",
       "      <td>2017-04-21</td>\n",
       "      <td>1</td>\n",
       "      <td>235.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2136283</th>\n",
       "      <td>154778256</td>\n",
       "      <td>0.0</td>\n",
       "      <td>420.0</td>\n",
       "      <td>2017-03-04</td>\n",
       "      <td>1</td>\n",
       "      <td>420.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3769027</th>\n",
       "      <td>168850880</td>\n",
       "      <td>6802744.0</td>\n",
       "      <td>184.0</td>\n",
       "      <td>2017-04-10</td>\n",
       "      <td>0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3127557</th>\n",
       "      <td>154771936</td>\n",
       "      <td>5334698.0</td>\n",
       "      <td>291.0</td>\n",
       "      <td>2017-03-04</td>\n",
       "      <td>1</td>\n",
       "      <td>290.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1945189</th>\n",
       "      <td>168899264</td>\n",
       "      <td>176292.0</td>\n",
       "      <td>334.0</td>\n",
       "      <td>2017-04-04</td>\n",
       "      <td>1</td>\n",
       "      <td>202.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1372132</th>\n",
       "      <td>168327168</td>\n",
       "      <td>912615.0</td>\n",
       "      <td>225.0</td>\n",
       "      <td>2017-04-03</td>\n",
       "      <td>0</td>\n",
       "      <td>93.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1183540</th>\n",
       "      <td>168043664</td>\n",
       "      <td>9842191.0</td>\n",
       "      <td>708.0</td>\n",
       "      <td>2017-04-03</td>\n",
       "      <td>0</td>\n",
       "      <td>708.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3200625</th>\n",
       "      <td>168162336</td>\n",
       "      <td>5262441.0</td>\n",
       "      <td>225.0</td>\n",
       "      <td>2017-04-23</td>\n",
       "      <td>1</td>\n",
       "      <td>186.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               uid    song_id  song_length        date  Churn  play_time  \\\n",
       "3179182  168559712   611618.0        235.0  2017-04-21      1      235.0   \n",
       "2136283  154778256        0.0        420.0  2017-03-04      1      420.0   \n",
       "3769027  168850880  6802744.0        184.0  2017-04-10      0        7.0   \n",
       "3127557  154771936  5334698.0        291.0  2017-03-04      1      290.0   \n",
       "1945189  168899264   176292.0        334.0  2017-04-04      1      202.0   \n",
       "1372132  168327168   912615.0        225.0  2017-04-03      0       93.0   \n",
       "1183540  168043664  9842191.0        708.0  2017-04-03      0      708.0   \n",
       "3200625  168162336  5262441.0        225.0  2017-04-23      1      186.0   \n",
       "\n",
       "        device  \n",
       "3179182     ar  \n",
       "2136283     ar  \n",
       "3769027   both  \n",
       "3127557   both  \n",
       "1945189     ar  \n",
       "1372132   both  \n",
       "1183540   both  \n",
       "3200625   both  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_play.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Timestamp('2017-04-28 00:00:00')"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_play['date'].max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "uid                     int64\n",
       "song_id               float64\n",
       "song_length           float64\n",
       "date           datetime64[ns]\n",
       "Churn                   int64\n",
       "play_time             float64\n",
       "device                 object\n",
       "dtype: object"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_play['date'] = pd.to_datetime(sample_play['date'])\n",
    "sample_play.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4235432, 7)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#original 4955455, after cleaning 4235432\n",
    "sample_play.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "How many active user: 4464\n",
      "How many churn user: 4334\n"
     ]
    }
   ],
   "source": [
    "#original 4465/4358, now 4464,4334\n",
    "print('How many active user:',len(sample_play.uid[sample_play.Churn == 0].unique()))\n",
    "print('How many churn user:',len(sample_play.uid[sample_play.Churn == 1].unique()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Slept for > 1 day\n"
     ]
    }
   ],
   "source": [
    "\n",
    "start = '2017-03-01'\n",
    "end = '2017-04-29'\n",
    "st,ed = pd.to_datetime(start), pd.to_datetime(end)\n",
    "elapsed = ed-st\n",
    "\n",
    "if elapsed > dt.timedelta(days = 1):\n",
    "    print(\"Slept for > 1 day\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "log_types = ['play', 'search', 'download']\n",
    "window_list = [1,3,7,14,21,30]\n",
    "snapshot_date = '2017-04-29'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_freq_avg(df, log_type, window, snapshot_date):\n",
    "    snapshot_date = pd.to_datetime(snapshot_date)\n",
    "    #df2 shows records inside window\n",
    "    df2 = df.loc[(snapshot_date - df['date'])<= dt.timedelta(window), :]\n",
    "    print('How many active/churn user after feature generation:',len(df2.uid[df2.Churn == 0].unique()), len(df2.uid[df2.Churn ==1].unique()))\n",
    "    df_feature = df2.groupby('uid')['play_time'] \\\n",
    "                    .agg(['count', 'mean']) \\\n",
    "                    .rename(columns = {'count':'freq_' + log_type + '_last_' + str(window) + '_days',\n",
    "                                       'mean' : 'avg_'+ log_type + 'time' + '_last_' + str(window) + '_days'})\n",
    "    #df_feature = df_feature.rename(columns = {'date':})\n",
    "    return df_feature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 143,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "How many active/churn user after feature generation: 2722 69\n"
     ]
    }
   ],
   "source": [
    "test1 = get_freq_avg(sample_play, 'play', window_list[0], snapshot_date)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Index(['freq_play_last_1_days', 'avg_playtime_last_1_days'], dtype='object') 2791\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>freq_play_last_1_days</th>\n",
       "      <th>avg_playtime_last_1_days</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>uid</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1901839.0</th>\n",
       "      <td>3</td>\n",
       "      <td>222.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2285308.0</th>\n",
       "      <td>11</td>\n",
       "      <td>107.818182</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5597959.0</th>\n",
       "      <td>6</td>\n",
       "      <td>111.166667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7533076.0</th>\n",
       "      <td>4</td>\n",
       "      <td>198.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7910960.0</th>\n",
       "      <td>14</td>\n",
       "      <td>214.142857</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           freq_play_last_1_days  avg_playtime_last_1_days\n",
       "uid                                                       \n",
       "1901839.0                      3                222.333333\n",
       "2285308.0                     11                107.818182\n",
       "5597959.0                      6                111.166667\n",
       "7533076.0                      4                198.000000\n",
       "7910960.0                     14                214.142857"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(test1.columns, test1.shape[0])\n",
    "test1.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_all_freq_avg(df, log_type, window_list, snapshot_time):\n",
    "    type_feature = get_freq_avg(df, log_type, window_list[0], snapshot_date)\n",
    "    for window in window_list[1:]:\n",
    "        #what = 'freq and average time of' + log_type + '_last_' + str(window) + '_days'\n",
    "        #print(\"Calculating\", what)\n",
    "        test_what = get_freq_avg(df, log_type, window, snapshot_date)\n",
    "        #print(test_what.head())\n",
    "        type_feature = pd.merge(test_what, type_feature, how='left', left_index=True, right_index=True)\n",
    "    return type_feature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "How many active/churn user after feature generation: 2714 65\n",
      "How many active/churn user after feature generation: 3616 180\n",
      "How many active/churn user after feature generation: 4080 382\n",
      "How many active/churn user after feature generation: 4317 656\n",
      "How many active/churn user after feature generation: 4393 880\n",
      "How many active/churn user after feature generation: 4461 1402\n"
     ]
    }
   ],
   "source": [
    "play_feature = get_all_freq_avg(sample_play, 'play', window_list, snapshot_date)\n",
    "#since many churn user never show up,  ┓( ´∀` )┏"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5863, 12)\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>freq_play_last_30_days</th>\n",
       "      <th>avg_playtime_last_30_days</th>\n",
       "      <th>freq_play_last_21_days</th>\n",
       "      <th>avg_playtime_last_21_days</th>\n",
       "      <th>freq_play_last_14_days</th>\n",
       "      <th>avg_playtime_last_14_days</th>\n",
       "      <th>freq_play_last_7_days</th>\n",
       "      <th>avg_playtime_last_7_days</th>\n",
       "      <th>freq_play_last_3_days</th>\n",
       "      <th>avg_playtime_last_3_days</th>\n",
       "      <th>freq_play_last_1_days</th>\n",
       "      <th>avg_playtime_last_1_days</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",
       "      <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>168087856</th>\n",
       "      <td>384</td>\n",
       "      <td>232.833333</td>\n",
       "      <td>143.0</td>\n",
       "      <td>215.097902</td>\n",
       "      <td>124.0</td>\n",
       "      <td>222.193548</td>\n",
       "      <td>26.0</td>\n",
       "      <td>193.461538</td>\n",
       "      <td>26.0</td>\n",
       "      <td>193.461538</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167558624</th>\n",
       "      <td>14</td>\n",
       "      <td>281.428571</td>\n",
       "      <td>14.0</td>\n",
       "      <td>281.428571</td>\n",
       "      <td>14.0</td>\n",
       "      <td>281.428571</td>\n",
       "      <td>14.0</td>\n",
       "      <td>281.428571</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167987504</th>\n",
       "      <td>313</td>\n",
       "      <td>157.440895</td>\n",
       "      <td>138.0</td>\n",
       "      <td>117.463768</td>\n",
       "      <td>36.0</td>\n",
       "      <td>185.583333</td>\n",
       "      <td>26.0</td>\n",
       "      <td>228.307692</td>\n",
       "      <td>23.0</td>\n",
       "      <td>229.000000</td>\n",
       "      <td>20.0</td>\n",
       "      <td>239.700000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168983872</th>\n",
       "      <td>106</td>\n",
       "      <td>219.990566</td>\n",
       "      <td>69.0</td>\n",
       "      <td>271.000000</td>\n",
       "      <td>69.0</td>\n",
       "      <td>271.000000</td>\n",
       "      <td>69.0</td>\n",
       "      <td>271.000000</td>\n",
       "      <td>69.0</td>\n",
       "      <td>271.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168309952</th>\n",
       "      <td>224</td>\n",
       "      <td>154.026786</td>\n",
       "      <td>119.0</td>\n",
       "      <td>155.974790</td>\n",
       "      <td>87.0</td>\n",
       "      <td>136.540230</td>\n",
       "      <td>85.0</td>\n",
       "      <td>136.623529</td>\n",
       "      <td>16.0</td>\n",
       "      <td>63.437500</td>\n",
       "      <td>11.0</td>\n",
       "      <td>65.454545</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167782720</th>\n",
       "      <td>81</td>\n",
       "      <td>173.506173</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>166351680</th>\n",
       "      <td>10</td>\n",
       "      <td>180.700000</td>\n",
       "      <td>10.0</td>\n",
       "      <td>180.700000</td>\n",
       "      <td>10.0</td>\n",
       "      <td>180.700000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75844624</th>\n",
       "      <td>34</td>\n",
       "      <td>237.617647</td>\n",
       "      <td>3.0</td>\n",
       "      <td>175.333333</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           freq_play_last_30_days  avg_playtime_last_30_days  \\\n",
       "uid                                                            \n",
       "168087856                     384                 232.833333   \n",
       "167558624                      14                 281.428571   \n",
       "167987504                     313                 157.440895   \n",
       "168983872                     106                 219.990566   \n",
       "168309952                     224                 154.026786   \n",
       "167782720                      81                 173.506173   \n",
       "166351680                      10                 180.700000   \n",
       "75844624                       34                 237.617647   \n",
       "\n",
       "           freq_play_last_21_days  avg_playtime_last_21_days  \\\n",
       "uid                                                            \n",
       "168087856                   143.0                 215.097902   \n",
       "167558624                    14.0                 281.428571   \n",
       "167987504                   138.0                 117.463768   \n",
       "168983872                    69.0                 271.000000   \n",
       "168309952                   119.0                 155.974790   \n",
       "167782720                     NaN                        NaN   \n",
       "166351680                    10.0                 180.700000   \n",
       "75844624                      3.0                 175.333333   \n",
       "\n",
       "           freq_play_last_14_days  avg_playtime_last_14_days  \\\n",
       "uid                                                            \n",
       "168087856                   124.0                 222.193548   \n",
       "167558624                    14.0                 281.428571   \n",
       "167987504                    36.0                 185.583333   \n",
       "168983872                    69.0                 271.000000   \n",
       "168309952                    87.0                 136.540230   \n",
       "167782720                     NaN                        NaN   \n",
       "166351680                    10.0                 180.700000   \n",
       "75844624                      NaN                        NaN   \n",
       "\n",
       "           freq_play_last_7_days  avg_playtime_last_7_days  \\\n",
       "uid                                                          \n",
       "168087856                   26.0                193.461538   \n",
       "167558624                   14.0                281.428571   \n",
       "167987504                   26.0                228.307692   \n",
       "168983872                   69.0                271.000000   \n",
       "168309952                   85.0                136.623529   \n",
       "167782720                    NaN                       NaN   \n",
       "166351680                    NaN                       NaN   \n",
       "75844624                     NaN                       NaN   \n",
       "\n",
       "           freq_play_last_3_days  avg_playtime_last_3_days  \\\n",
       "uid                                                          \n",
       "168087856                   26.0                193.461538   \n",
       "167558624                    NaN                       NaN   \n",
       "167987504                   23.0                229.000000   \n",
       "168983872                   69.0                271.000000   \n",
       "168309952                   16.0                 63.437500   \n",
       "167782720                    NaN                       NaN   \n",
       "166351680                    NaN                       NaN   \n",
       "75844624                     NaN                       NaN   \n",
       "\n",
       "           freq_play_last_1_days  avg_playtime_last_1_days  \n",
       "uid                                                         \n",
       "168087856                    NaN                       NaN  \n",
       "167558624                    NaN                       NaN  \n",
       "167987504                   20.0                239.700000  \n",
       "168983872                    NaN                       NaN  \n",
       "168309952                   11.0                 65.454545  \n",
       "167782720                    NaN                       NaN  \n",
       "166351680                    NaN                       NaN  \n",
       "75844624                     NaN                       NaN  "
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(play_feature.shape)\n",
    "play_feature.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "# add dropped churn user back\n",
    "#sample_uid_8798.shape\n",
    "play_feature_8798 = play_feature.join(sample_uid_8798, how = 'outer')\n",
    "                                "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(8798, 13)\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>freq_play_last_30_days</th>\n",
       "      <th>avg_playtime_last_30_days</th>\n",
       "      <th>freq_play_last_21_days</th>\n",
       "      <th>avg_playtime_last_21_days</th>\n",
       "      <th>freq_play_last_14_days</th>\n",
       "      <th>avg_playtime_last_14_days</th>\n",
       "      <th>freq_play_last_7_days</th>\n",
       "      <th>avg_playtime_last_7_days</th>\n",
       "      <th>freq_play_last_3_days</th>\n",
       "      <th>avg_playtime_last_3_days</th>\n",
       "      <th>freq_play_last_1_days</th>\n",
       "      <th>avg_playtime_last_1_days</th>\n",
       "      <th>Churn</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",
       "      <th></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>167831584</th>\n",
       "      <td>2588.0</td>\n",
       "      <td>246.847759</td>\n",
       "      <td>2325.0</td>\n",
       "      <td>248.389677</td>\n",
       "      <td>1758.0</td>\n",
       "      <td>247.850398</td>\n",
       "      <td>818.0</td>\n",
       "      <td>236.882641</td>\n",
       "      <td>287.0</td>\n",
       "      <td>226.756098</td>\n",
       "      <td>169.0</td>\n",
       "      <td>210.911243</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168791952</th>\n",
       "      <td>1894.0</td>\n",
       "      <td>139.876729</td>\n",
       "      <td>1098.0</td>\n",
       "      <td>125.788274</td>\n",
       "      <td>766.0</td>\n",
       "      <td>133.734366</td>\n",
       "      <td>271.0</td>\n",
       "      <td>146.929889</td>\n",
       "      <td>202.0</td>\n",
       "      <td>142.762376</td>\n",
       "      <td>32.0</td>\n",
       "      <td>92.656250</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168825856</th>\n",
       "      <td>1306.0</td>\n",
       "      <td>169.735069</td>\n",
       "      <td>526.0</td>\n",
       "      <td>162.684411</td>\n",
       "      <td>174.0</td>\n",
       "      <td>158.758621</td>\n",
       "      <td>76.0</td>\n",
       "      <td>170.157895</td>\n",
       "      <td>34.0</td>\n",
       "      <td>195.147059</td>\n",
       "      <td>11.0</td>\n",
       "      <td>258.636364</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36567948</th>\n",
       "      <td>7.0</td>\n",
       "      <td>122.142857</td>\n",
       "      <td>7.0</td>\n",
       "      <td>122.142857</td>\n",
       "      <td>7.0</td>\n",
       "      <td>122.142857</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168039056</th>\n",
       "      <td>1111.0</td>\n",
       "      <td>179.917192</td>\n",
       "      <td>517.0</td>\n",
       "      <td>209.930368</td>\n",
       "      <td>197.0</td>\n",
       "      <td>231.365482</td>\n",
       "      <td>133.0</td>\n",
       "      <td>248.849624</td>\n",
       "      <td>132.0</td>\n",
       "      <td>249.106061</td>\n",
       "      <td>13.0</td>\n",
       "      <td>219.615385</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168869312</th>\n",
       "      <td>360.0</td>\n",
       "      <td>151.663889</td>\n",
       "      <td>120.0</td>\n",
       "      <td>200.350000</td>\n",
       "      <td>101.0</td>\n",
       "      <td>200.049505</td>\n",
       "      <td>11.0</td>\n",
       "      <td>236.818182</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154597840</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154293776</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           freq_play_last_30_days  avg_playtime_last_30_days  \\\n",
       "uid                                                            \n",
       "167831584                  2588.0                 246.847759   \n",
       "168791952                  1894.0                 139.876729   \n",
       "168825856                  1306.0                 169.735069   \n",
       "36567948                      7.0                 122.142857   \n",
       "168039056                  1111.0                 179.917192   \n",
       "168869312                   360.0                 151.663889   \n",
       "154597840                     NaN                        NaN   \n",
       "154293776                     NaN                        NaN   \n",
       "\n",
       "           freq_play_last_21_days  avg_playtime_last_21_days  \\\n",
       "uid                                                            \n",
       "167831584                  2325.0                 248.389677   \n",
       "168791952                  1098.0                 125.788274   \n",
       "168825856                   526.0                 162.684411   \n",
       "36567948                      7.0                 122.142857   \n",
       "168039056                   517.0                 209.930368   \n",
       "168869312                   120.0                 200.350000   \n",
       "154597840                     NaN                        NaN   \n",
       "154293776                     NaN                        NaN   \n",
       "\n",
       "           freq_play_last_14_days  avg_playtime_last_14_days  \\\n",
       "uid                                                            \n",
       "167831584                  1758.0                 247.850398   \n",
       "168791952                   766.0                 133.734366   \n",
       "168825856                   174.0                 158.758621   \n",
       "36567948                      7.0                 122.142857   \n",
       "168039056                   197.0                 231.365482   \n",
       "168869312                   101.0                 200.049505   \n",
       "154597840                     NaN                        NaN   \n",
       "154293776                     NaN                        NaN   \n",
       "\n",
       "           freq_play_last_7_days  avg_playtime_last_7_days  \\\n",
       "uid                                                          \n",
       "167831584                  818.0                236.882641   \n",
       "168791952                  271.0                146.929889   \n",
       "168825856                   76.0                170.157895   \n",
       "36567948                     NaN                       NaN   \n",
       "168039056                  133.0                248.849624   \n",
       "168869312                   11.0                236.818182   \n",
       "154597840                    NaN                       NaN   \n",
       "154293776                    NaN                       NaN   \n",
       "\n",
       "           freq_play_last_3_days  avg_playtime_last_3_days  \\\n",
       "uid                                                          \n",
       "167831584                  287.0                226.756098   \n",
       "168791952                  202.0                142.762376   \n",
       "168825856                   34.0                195.147059   \n",
       "36567948                     NaN                       NaN   \n",
       "168039056                  132.0                249.106061   \n",
       "168869312                    NaN                       NaN   \n",
       "154597840                    NaN                       NaN   \n",
       "154293776                    NaN                       NaN   \n",
       "\n",
       "           freq_play_last_1_days  avg_playtime_last_1_days  Churn  \n",
       "uid                                                                \n",
       "167831584                  169.0                210.911243      0  \n",
       "168791952                   32.0                 92.656250      0  \n",
       "168825856                   11.0                258.636364      0  \n",
       "36567948                     NaN                       NaN      1  \n",
       "168039056                   13.0                219.615385      0  \n",
       "168869312                    NaN                       NaN      0  \n",
       "154597840                    NaN                       NaN      1  \n",
       "154293776                    NaN                       NaN      1  "
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(play_feature_8798.shape)\n",
    "play_feature_8798.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "8798 5273 8823\n"
     ]
    }
   ],
   "source": [
    "print(len(sample_play.uid.unique()), play_feature.shape[0], play_feature_8823.shape[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x3412e70>"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x34252f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plot freq_play_x_days in one figure..not clear enough, separate them\n",
    "\n",
    "window_list = [1,3,7,14,21,30]    \n",
    "fig, ax = plt.subplots()\n",
    "legend_list = []\n",
    "for w in window_list:\n",
    "    colname = 'freq_play_last_' + str(w) + '_days'\n",
    "    legend_list.append(colname)\n",
    "    play_feature_8798[colname].hist(bins =30, alpha = 0.5)\n",
    "    \n",
    "plt.title('Histogram of frequency of different play_time period' , fontsize = 13)\n",
    "plt.xlabel(\"Frequency of play time\", fontsize = 13)\n",
    "plt.legend(legend_list, loc='upper right', fontsize = 13)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0\n",
      "1\n",
      "2\n",
      "3\n",
      "4\n"
     ]
    }
   ],
   "source": [
    "for w in window_list:\n",
    "    i = window_list.index(w)\n",
    "    print(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x3590690>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plot separately\n",
    "f, axs = plt.subplots(1,6, sharex = True, sharey=True, figsize = (20,5))\n",
    "for w in window_list:\n",
    "    i = window_list.index(w)\n",
    "    colname = 'freq_play_last_' + str(w) + '_days'\n",
    "    plt.xlabel(\"frequency of play_time\")\n",
    "    plt.ylabel('count')\n",
    "    play_feature_8798[colname].hist(ax = axs[i], bins = 30)\n",
    "\n",
    "    axs[i].set_title(colname)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x3205d90>"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x3205b10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plot average play time in x days\n",
    "fig, ax = plt.subplots()\n",
    "legend_list = []\n",
    "for w in window_list:\n",
    "    colname = 'avg_playtime_last_' + str(w) + '_days'\n",
    "    legend_list.append(colname)\n",
    "    play_feature_8798[colname].hist(bins =30, alpha = 0.5)\n",
    "    \n",
    "plt.title('Histogram of average playtime of different play_time period' , fontsize = 13)\n",
    "plt.xlabel(\"Average of play time\", fontsize = 13)\n",
    "plt.legend(legend_list, loc='upper right', fontsize = 13)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xeb223d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "f, axs = plt.subplots(1,6, sharex = True, sharey=True, figsize = (20,5))\n",
    "for w in window_list:\n",
    "    i = window_list.index(w)\n",
    "    colname = 'avg_playtime_last_' + str(w) + '_days'\n",
    "    plt.xlabel(\"average play time\")\n",
    "    plt.ylabel('count')\n",
    "    play_feature_8798[colname].hist(ax = axs[i], bins = 30)\n",
    "\n",
    "    axs[i].set_title(colname)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###  1.2. device feature\n",
    "## Undate by 318, so after that, no need to create device feature, can directly go search/download feature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(12618, 3)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_play[['uid', 'device', 'Churn']].drop_duplicates().shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8823, 2)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_play[['uid', 'Churn']].drop_duplicates().shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Adding device as another groupby condition causing significant increase, we can assume that some users have MusicBox installed in more than one device, such as their android phone and ipad.\n",
    "\n",
    "Create a new value for user who have both device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "device_count = sample_play[['uid', 'device']].drop_duplicates() \\\n",
    "                                              .groupby('uid')['device'] \\\n",
    "                                              .agg(['max', 'count'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "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>max</th>\n",
       "      <th>count</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>uid</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>400904.0</th>\n",
       "      <td>ar</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>838955.0</th>\n",
       "      <td>ar</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>971787.0</th>\n",
       "      <td>ar</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1264383.0</th>\n",
       "      <td>ar</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1901839.0</th>\n",
       "      <td>ar</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          max  count\n",
       "uid                 \n",
       "400904.0   ar      1\n",
       "838955.0   ar      1\n",
       "971787.0   ar      1\n",
       "1264383.0  ar      1\n",
       "1901839.0  ar      1"
      ]
     },
     "execution_count": 100,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "device_count.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "device_count['device'] = np.where(device_count['count'] == 1, device_count['max'], 'both')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ar      4885\n",
       "both    3751\n",
       "ip       162\n",
       "Name: device, dtype: int64"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "device_count['device'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As we assumed, there are 40% user have both device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "device_feature = sample_play[['uid', 'Churn']].drop_duplicates() \\\n",
    "                                               .reset_index(drop=True) \\\n",
    "                                               .set_index('uid') \\\n",
    "                                               .join(device_count['device'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(8798, 2)\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>Churn</th>\n",
       "      <th>device</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>uid</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>167843536</th>\n",
       "      <td>0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154483216</th>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154606848</th>\n",
       "      <td>1</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168612336</th>\n",
       "      <td>1</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167972800</th>\n",
       "      <td>0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167881776</th>\n",
       "      <td>0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167936640</th>\n",
       "      <td>0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>156894672</th>\n",
       "      <td>0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           Churn device\n",
       "uid                    \n",
       "167843536      0     ar\n",
       "154483216      1     ar\n",
       "154606848      1   both\n",
       "168612336      1   both\n",
       "167972800      0   both\n",
       "167881776      0   both\n",
       "167936640      0   both\n",
       "156894672      0     ar"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(device_feature.shape)\n",
    "device_feature.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8798"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#device_feature = device_feature.reset_index(inplace = False)\n",
    "device_feature.groupby(['device', 'Churn'])['uid'].count().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Churn      0     1\n",
      "device            \n",
      "ar      2185  2700\n",
      "both    2256  1495\n",
      "ip        23   139\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x33c9b50>"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZ0AAAExCAYAAACnAX83AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAIABJREFUeJzt3Xm4HFWd//H3h5Cwh8UQAnFCQBwMBAQSkLBEwEnYVBz0GZVFIioqI+ogw+P20+AuILKIbOoEEJRHcWETCArCEHBMEFlFkTUJ2VgCWQghfH9/nNOkUum+t+9N3+qbez+v5+nn3qpz6pxT1dX17VPndLciAjMzsyqs0+4GmJlZ/+GgY2ZmlXHQMTOzyjjomJlZZRx0zMysMg46ZmZWGQedLpB0m6Qf9EC5IyWFpLF5+YC8PKTVdeXye2Q/ukPSCZKekvSapMlrUE6PHrPepnzO9GA9G0r6paSFub6RPVlfq7XyvJD0wJqcow3KDEnva2WZvV2/DzqSpuQnPiQtlzRP0q2S/lPSwFL2I4EvNFnuZEkPNNmMp4GtgXu70PRm2jBJ0qI6SU3vR0+StDlwPnAGMBw4s4O8u0m6StIcSS9LejQ/d7tU1d5mSbpQ0vdL606WtELSN7tRXr03CT1yztRxPDAe2C/X93QP19ffbA1c21OFd3ANaJt+H3SyW0hP/khgIukkOA24Q9JGtUwR8VxEvNTKiiUNiogVETEnIl5tZdmN9MR+dNO2wLrAdRHxTETUfXFIeifwJ2Bj4FhgFPAB4BngOz3ZQEnrSBrQhfwC3gX8tpT0EVJbJ3WlvEYqPGd2AB6OiPtzfSu6WkBXj2F/ko/psna3o1IR0a8fwBTSRa+8fjTwCnBaYd1twA8Ky0cC9wFLgeeAPwJbAZOAKD0m5W0C+E/gV8Bi0rv7kXn92JzngLz8TtI72ZeBGcCYQt2TgEWlNte2G1L4v/iY3GA/NgcuBZ7P+3ILsHO5LuAdwAO53bcC23VybEcAvwZeyo9fAW8slFlu38g6ZWwIzAeuaVDHZqV9fwcpQC0BpgN7NHvMSvt6WN7XV/O5MAW4DvgMMCsfq/8BNiyVtxfwLLBuYd04YC4wEHgUeGed/dgb+EM+tguB3wPb5HpXO04UzhnSm8eZwEmlMv8159k9L28KXAzMy8/HH8nnXINje1up3tu6eL6scgwb1DEc+Hku63ngeuDNhfQ3kQL4nHxs7ikfP2AQ8C3gSWAZ8Bjw6WbPiwbtGprrXZrLPT7vy+RCnobHM6ctBd5VKncisBwYWrgevK+Qvg1wRT6HlpBe/wcW0t9Fuha8DDwOfBMY1GAfavu+yjUA+ArwQJ38dwLnFq+LwJdJ5+4i0vm+QSG/gFOBf+Z9vR84ptNrbncu1H3pQYOgk9OuKT45FC7WwDBSUPoc6QIwGvgoKehsQAomf8v5htWerPzEz8t5twe2o3HQ+RtwcC77F6QX3obFF3aDk2wI6YX4GdILtdaGjcv7kZd/m+saD+yS9/vpQpsn5RfKLaSL6q7AX4CbOjiuIl0gpgF7ki6Od5Ne8MrH6ODc3j1z+wbUKeffc559Onkea/v+f8CBwFuAm4CHATVzzAp5Xs3t3pd04d4knycLgUtIPa2JwAvAF0rlfQu4rLTux8A5+f/JwG9K6W8lvWgvBnbL5X+cFLQ3zW35SeF5HFDnnDkDuLtU7mnAg4Xn439JF/W9SD2YrwMvAls3OKZb5Hqn5Xq36ML5stoxrFP+hsDf87HdNT9nPyJd5DcsHJtP5Hp2AL5Eet29pVDOz0hB972k19SBwIeaPS8a7PsNwIO5/buTXjOLWPnGrdPjSXrN/rxU7qXA9YXl14MOsBHwD9LFfzwp4B5JDjqk18uLwIdz2oHAI8CZDfah7jUAeGN+fvYq5N0xt+WtheviS3kfRue6Z5GDUs7zzVz/IaTr2FG5rsM7fK1WdXHvrQ86DjrfAZYUlm9jZdDZIz9J2zbYdjL1300EcF5p3UjqB52jC3k2Jl3kPhorX9hNvWuv04bifrw5bzO+kL4p6QJbrCuAHQt5jia9+NdpsP8TgBUUei+kC8JrwL/l5bE06OEUtjk159m8k+extu8HF9btm9cVe1fNHLOg0KssnCdPs2oP5hLgllK+B4EjS8/bS4Xndvt83IYV8lxBKWA0er46OGd2zcs7FPL8gxwUgYNIF80NSuXcC5zaQd0/IPdwunG+jGlUbs53fG6jCusGkN7l/0cH290NfLnUnkO6e17U2abWQ9y3sG7bfD5PbvZ4AkeQeiub5OUNSEHjg6XrQS3ofCyfK0MatOt24P+V1r0nt6NuAKXxNeA64MLC8neB6aXz/QXyG9W87hhST3Kj/FgK7F8q92zgho6ed4/pdEykk6Kev5Le+T8g6WpJn5S0ZZPlTm8y3121fyKNd9wP7NTkts0aRQoExboW1qlrWUQ8UlieTbpdtFkH5c6OiCcK5T6Wt+vKPqgLeSHd7iy2EdKtkq54lfoD9A/FqmMos4tlS9qBFFRuKuT5ADAzIqbD68fgz8BxhTy7k26ndVtE3Ed6zo7KbXkb6d3wlTnLGPKtSkmLag/Su9g3daGqZs+XRsewaAzpHfJLhfYsJN2+e1Pej40knS7pIUnP5zxjSb1ASMfuNdLt3o505byo7eP/1VZExJOF7Wpt7+x43kAKOv+el99NOp/L4301uwP3RcSCBuljgC+V6ruSFACGNdimkUuAD0jaII+3HUvqkRfdF6uOs95F6j29ifRcrw/cWGrPJ+nkfFq3iw3tb3Yi3R9eTUSskDSRdC9+Immg+NuS3h4Rf+2k3MUtaNtrrH5BLs+2a0ZHF/ViwC0PWNfSGr1x6ShgN1pfz9/z31Gk2zWdWV6nnlobmz1my6L+gPny0nKw6v6/B/h9RBSf348CO0oqHr91gC1J7y6p06buuoLUe/gaqSd6R75Y1uqcC+xfZ7sXu1BHs+dLo2NYtA4pMH2gTtpz+e+ZpNs3p5B6RUuAy0gXv87aU9TReVHWTJmdHs+IWC7pF6Tn4rL891cRsaSb9a5DumX6izpp85toc9H1pGP5XlKg34x0m7JZtWP3LuCpUlr5dVJ3QyuRNJp0sv+yUZ5I7oqI00jjErOB9+fkV0i3CtbE3oX2bER6F/VwXjUf2FDS4EL+3UrbN9OGh0jnwbhCXYNJ99Af6l6zXy93ePFzHZK2Jw2UdqXcm4EFwOfrJUpq1NOqp5ljtiaOAH5TW5C0M/A20puS3QqPtwEjJY3PWe8h3a5ppNlz6QpgB0l7k87DnxbS7iGNN74WEY+WHvOa2ruklefLPaSxkAV12lQLOvuRxsiuzr25maz6Tvqe3J4Du1h3Rx7OZe5ZWyFpBOncLdbbzPH8KfAOSTuRrifF56TsHmDXDj5TdA9pLKtc36PReBZj3XMn559CepNyPCkYvlDKtktx9i7pevQKaeLAQ6RbbdvWacuTdKSje2/94ZEP/FRS93Qb0sDlyaQL3d3ARoW8t7FyLGRv0syOPUld/SNI92OPyelHkd5J7EEa2F8vSvdwC+WOpP6YzkOksZGdgatI76w2ynm2IN3LPZ/0wn0vaTZLcXxin7w8Ibdhw/J+5OXfkF5o+9N4YLjDsZA6x7U2keBO0m2BsaTu+XRWDux3OqaT8x1BOtmvz/syMh/Xr5MHZeu1p85xbeaYrbavhfPkutK6yeRxO1LPZTmwVSH9+8A9Dfbpd+QJB6RA9DJpIsFbSYO6HwVG5PSLSTOWRubncZ3yvhXKvZWVMx43Kz0fd5Bugx1Kuq01jvTOef96bczbrTKm093zpUHZG5IGov8IvD23aTzwPfIMNuBq0q2xPXJdvyS9M59SKOcqVk4k2C6369hmz4sOnp/78zHajTSzsDyRoKnjSbpI30t6UzqglFYc09ko5/3fvA/bkW7JFScSLCf1ZEeTJkW8Dzi9g/2oew3IaduTxqmWU5ghVzjfX8rHdue8/dOset34Bmn87XjS62k30qSPEzp83rtyge6LD1adkvoqKdjcBpxEaSoiqwadUfnEnEuK+I9SGJAF1ssvkOdZfcp0s0Hn3aQX3DLSBXzP0nZHkG4/LSWNIxxT5wV2Qd6nYA2nTJfqPqBcV51jO4J0gapNmf41hcFbmgw6Oe8Y0m2F2vH+Z37udm7UnvJxbeaY1dvXwnnSUdA5HphWSBtE6ll9scH+HE96U7JpXt6PNFC8lDSAewsrZ0H9KylgL6kdr3r7Vig3gKvr1LkJcA7pAv0K6SLyc+BNHRz3ekGny+dLB+VvRZqKOy8/r4+TZszVno9tc/mLc7tPIQ2CTym91k4nza6qnRuf6sp50aBd1+T9e5r0JqA8Zbqp40kKEgF8r049q1wPSDPLrsrnwBLSLNEDCukTScFuCek23vTavnawL6tdAwppf8jHS6X1U/Jx/kp+bhbl57wYtES6TtZ6PfNJb+AndNSe2jtOM1sDkn4L3BkRp7e7LWbNkvQQcEVEfLO0fgopUL+z1XV6IoFZa9xJ1wZizdpG0lDgg6Re30VV1u2gY9YC7uHYWmYu6Zbbx6PxFO0e4dtrZmZWGU+ZNjOzyjjomJlZZTymUzBkyJAYOXJku5thZrZWmTFjxoKIaOprwBx0CkaOHMn06c1+LZqZmQFI6vhbCAp8e83MzCrjoGNmZpVx0DEzs8o46JiZWWUcdMzMrDKevWZm1k2vvfYaCxYs4IUXXmDFis5+s27tNGDAADbbbDOGDBnCOuuseT/FQcfMrJtmzpyJJEaOHMnAgQORWvUjsL1DRLB8+XLmzp3LzJkzGTFiROcbdcK318zMumnx4sUMHz6cQYMG9bmAAyCJQYMGMXz4cBYvXtz5Bk1wT6c/mbxpxfUtrLY+szZoxS2n3q6V+9j3j5aZmfUaDjpmZlYZ314zM2uxkZ+/vtL6nvjO4ZXWtybc0zEz66d++MMfst1227H++uszZswY7rjjjh6v00HHzKwfuuqqq/jMZz7DF7/4Rf7yl7+wzz77cOihh/LUU0/1aL0OOmZm/dBZZ53FpEmT+NjHPsaoUaM477zz2Hrrrbngggt6tF4HHTOzfuaVV15hxowZTJw4cZX1EydOZNq0aT1at4OOmVk/s2DBAlasWMFWW221yvqtttqKOXPm9GjdDjpmZv1U+VsUIqLHv1nBQcfMrJ8ZMmQIAwYMWK1XM2/evNV6P61WWdCR9AVJf5b0oqT5kq6VNLqUZ4qkKD3uLuVZT9J5khZIWizpGklvLOUZkctfnPOdK2lQFftpZtbbDRo0iDFjxjB16tRV1k+dOpV99tmnR+uu8sOhBwA/BP4MCPgacIuknSLiuUK+W4BjC8uvlMo5GzgC+CDwLHAWcJ2kMRGxQtIA4Pqctj/wBuDSXOdJrd4pM7O10cknn8yxxx7LXnvtxb777suFF17I7Nmz+cQnPtGj9VYWdCLi4OKypGOBhcC+wLWFpGURUXckS9KmwEeAD0fE1EI5TwL/BtwETAR2BraNiKdznlOBH0n6UkS82NIdMzMrWRu+IeD9738/zz77LN/4xjd45plnGD16NDfccAPbbrttj9bbzjGdTXL9z5fW7ydpnqS/S7pE0tBC2hhgIHBzbUUOLA8DtT7hOODhWsDJbgLWy9ubmRlw4okn8sQTT7Bs2TJmzJjB+PHje7zOdgadc4B7gbsK624EPgS8A/gcsBfwB0nr5fRhwApgQamsuTmtlmduKX1B3m5YaT2STpA0XdL0+fPnd39vzMysU235wk9JZwH7AftFxOu/8RoRPy9ku1/SDNKts8OBX3VUJBCF5WiQb7X1EXExcDHA2LFjG21nZmYtUHlPR9L3SZMADoqIxzrKGxGzgZnAm/OqOcAAYEgp61BW9m7msHqPZkjertwDMjOzClUadCSdAxxFCjh/ayL/EGA48ExeNQNYDkwo5HkjMAqofXfDXcCo0jTqCcCyvL2ZmbVJZbfXJJ1Pmgr9HuB5SbXeyKKIWCRpY2AycDUpyIwEvg3MA34NEBELJf0YOEPSPFZOmb6PNNUa0iSDB4HLJH2ONGX6DOASz1wzM2uvKns6J5JmrP2eFFRqj1Ny+gpgF+C3wN9Jn615BBgXES8Vyvkv0vjOVcCdwCLgXbWxofz3cGBJTr8q5z8FMzNrqyo/p9PhF/pExFLg4I7y5Hwvkz7k2fCDnhHxFPDOrrbRzMx6lr97zczMKuOgY2ZmlWnL53TMzPq0yZtWXN/CautbA+7pmJn1M7fffjvvfve7GT58OJKYMmVKZXU76JiZ9TOLFi1i9OjRnHPOOWywwQaV1u3ba2Zm/cxhhx3GYYcdBsCkSZMqrds9HTMzq4yDjpmZVcZBx8zMKuOgY2ZmlXHQMTOzynj2mplZP7No0SIeffRRAF577TWeeuop7r33XrbYYgtGjBjRo3U76Jh108jPX19pfU985/BK67M10Mu/IWD69OkceOCBry9/9atf5atf/SrHHXdcj39Q1EHHzKyfOeCAA4iIttTtMR0zM6uMg46ZmVXGQcfMzCrjMZ02qnwgev1KqzMzW417OmZma6BdA/JVauU+OuiYmXXTwIEDWbp0abub0eOWLl3KwIEDW1KWg46ZWTcNHTqUWbNmsWTJkj7Z44kIlixZwqxZsxg6dGhLyvSYjplZNw0ePBiA2bNns3z58ja3pmcMHDiQrbba6vV9XVMOOmZma2Dw4MEtuyD3B769ZmZmlXHQMTOzyjjomJlZZRx0zMysMg46ZmZWmcqCjqQvSPqzpBclzZd0raTRpTySNFnSbElLJd0maedSns0lXS5pYX5cLmmzUp5dJP0xlzFL0lckqYr9NDOzxqrs6RwA/BDYBzgIeBW4RdIWhTynAp8DTgL2BOYBUyVtUshzJbAHcChwSP7/8lqipMHAVGBuLuPTwH8DJ/fETpmZWfMq+5xORBxcXJZ0LLAQ2Be4NvdEPgt8JyKuznmOIwWeo4CLJI0iBZr9ImJazvNx4A5JO0bEI8DRwIbAcRGxFHggb3eypLOiL35s2MxsLdHOMZ1Ncv3P5+XtgGHAzbUMOWjcTuodAYwDFgHTCuXcCSwu5bkjb1tzE7ANMLKle2BmZl3SzqBzDnAvcFdeHpb/zi3lm1tIGwbML/ZW8v/zSnnqlVGs43WSTpA0XdL0+fPnd2c/zMysSW0JOpLOAvYD3hsRK0rJ5dtfKq2rd3usszxqsJ6IuDgixkbE2C233LLTtpuZWfdVHnQkfR/4IHBQRDxWSJqT/5Z7I0NZ2VOZAwwtzkTL/29ZylOvDFi9B2RmZhWqNOhIOoc0KeCgiPhbKflxUsCYUMi/PrA/K8dw7gI2Jo3b1IwDNirl2T9vWzMBmA080ZIdMTOzbqnyczrnAx8m9XKelzQsPzaG18dmzgY+L+nI/BmeKaSJA1fmPA8DN5Jmsu0taRxwEXBdnrlGzrsEmCJptKQjgc8DnrlmZtZmVf60wYn57+9L608DJuf/Twc2AM4HNgf+BEyMiJcK+Y8GzmXlLLdrgE/VEiNioaQJuYzppNlx3wPOatWOmJlZ91T5OZ1OvxEg90QmszII1cvzHHBMJ+XcD4zvWgvNzKyn+bvXzMysMg46ZmZWGQcdMzOrjIOOmZlVxkHHzMwq46BjZmaVcdAxM7PKOOiYmVllHHTMzKwyDjpmZlYZBx0zM6uMg46ZmVXGQcfMzCrjoGNmZpVx0DEzs8o46JiZWWUcdMzMrDIOOmZmVhkHHTMzq0zTQUfSeEnr1lm/rqTxrW2WmZn1RV3p6dwKbFFn/aY5zczMrENdCToCos76NwCLW9McMzPry1a7XVYm6Zr8bwA/lbSskDwAGA1M64G2mZlZH9Np0AGezX8FPA8sLaS9AvwvcEmL22VmZn1Qp0EnIj4MIOkJ4MyI8K00MzPrlmZ6OgBExGk92RAzM+v7mg46krYAvgm8AxhKaRJCRAxubdPMzKyv6crstR8DhwJTgM8CJ5Uencqf9blG0ixJIWlSKX1KXl983F3Ks56k8yQtkLQ4l/fGUp4Rkq7N6QsknStpUBf21czMekDTPR1SD2dCRPxpDerbGHgAuCw/6rkFOLaw/Eop/WzgCOCDpEkOZwHXSRoTESskDQCuz2n7k6Z0X0qaCNFUcDQzs57RlaAzD1i0JpVFxA3ADZB6NQ2yLYuIOfUSJG0KfAT4cERMzeuOBZ4E/g24CZgI7AxsGxFP5zynAj+S9KWIeHFN9sHMzLqvK7fXvgR8TdLGPdWYbD9J8yT9XdIlkoYW0sYAA4GbaytyYHkY2CevGgc8XAs42U3Aenl7MzNrk670dL4MjATmSXoSWF5MjIhdW9CeG4FfAY/nur4B/CHfOlsGDANWAAtK283NaeS/c0vpC/J2w0rrkXQCcALAiBEjWrALZj1k8qYV17ew2vqsX+hK0Pllj7Uii4ifFxbvlzSDdOvscFIwaqT8FT31vq6n7vqIuBi4GGDs2LGNtjMzsxbo1Z/TiYjZkmYCb86r5pC+emcIML+QdShweyHPvqWihuTtyj0gMzOrUK/+PR1JQ4DhwDN51QzSbb0JhTxvBEax8vvf7gJGlaZRTwCW5e3NzKxNuvLh0JdofNuqqQ+H5kkIO+TFdYARknYDnsuPycDVpCAzEvg2adbcr3MdCyX9GDhD0jxWTpm+jzTVGtIkgweByyR9jjRl+gzgEs9cMzNrr66M6XyqtDwQ2B14L+mbCpoxllV/e+e0/LgU+CSwC/AhYDNS4LkV+I+IeKmwzX8BrwJXARsAvwc+FBErAPJndQ4HfgjcSfqC0iuBU5pso5mZ9ZCujOlcWm+9pHtIHxw9r4kybiMN+jdycBNlvEwn34IQEU8B7+ysLDMzq1YrxnRuBd7VgnLMzKyPa0XQ+QCrf27GzMxsNV2ZSHA/q04kELAVsAVpPMbMzKxDa/Lh0NdIn5W5LSL+1rommZlZX9WrPxxqZmZ9S1d6OgBIOgjYiXSr7cE8I83MzKxTXRnTGU76kOYYYHZevY2k6cC/R8TshhubmZnRtdlr55K+qXmHiPiXiPgX0neirchpZmZmHerK7bUJwAER8XhtRUQ8JunTpG8FMDMz61ArPqfzWgvKMDOzfqArQef3wLmS/qW2QtII4Bzc0zEzsyZ0Jeh8GtgQeEzSk5KeAP6Z1326B9pmZmZ9TFc+p/M0sIekCcBbSN9I8FBE3NLxlmZmZkmnPR1Jh0p6QtKmABExNSLOi4hzgT/ntIk93lIzM1vrNXN77VPAGRGxsJyQ130X+EyrG2ZmZn1PM0FnV1b+Kmc9fwDe2prmmJlZX9ZM0NmSjqdFB+knoc3MzDrUTNCZSertNLIrMKs1zTEzs76smaBzPfB1SRuUEyRtCHwt5zEzM+tQM1Omvwm8D/iHpPOA2m/njCJNMhDwrZ5pnpmZ9SWdBp2ImCdpH+ACUnBRLQm4CTgxIub2XBPNzKyvaOrDoRHxJHCYpM2BHUiB5x8R8XxPNs7MzPqWLv2IWw4yf+6htpiZWR/Xim+ZNjMza4qDjpmZVcZBx8zMKuOgY2Zmlak06EgaL+kaSbMkhaRJpXRJmixptqSlkm6TtHMpz+aSLpe0MD8ul7RZKc8ukv6Yy5gl6SuShJmZtVXVPZ2NgQdI30q9tE76qcDngJOAPYF5wFRJmxTyXAnsARwKHJL/v7yWKGkwMBWYm8v4NPDfwMkt3hczM+uiLk2ZXlMRcQNwA4CkKcW03BP5LPCdiLg6rzuOFHiOAi6SNIoUaPaLiGk5z8eBOyTtGBGPAEeTfs30uIhYCjyQtztZ0lkRERXsqpmZ1dGbxnS2A4YBN9dW5KBxO7BPXjUOWARMK2x3J7C4lOeOvG3NTcA2wMieaLiZmTWnNwWdYflv+St15hbShgHzi72V/P+8Up56ZRTreJ2kEyRNlzR9/vz5a9B8MzPrTG8KOjXl218qrat3e6yzPGqwnoi4OCLGRsTYLbfcsqttNTOzLuhNQWdO/lvujQxlZU9lDjC0OBMt/79lKU+9MmD1HpCZmVWoNwWdx0kBY0JthaT1gf1ZOYZzF2kG3LjCduOAjUp59s/b1kwAZgNP9ETDzcysOVV/TmdjSbtJ2i3XPSIvj8hjM2cDn5d0pKTRwBTSxIErASLiYeBG0ky2vSWNAy4Crssz18h5lwBTJI2WdCTwecAz18zM2qzqns5Y4C/5sQFwWv7/azn9dOAs4HxgOrA1MDEiXiqUcTTwV9Ist5vy/8fWEiNiIalns00u43zge7lcMzNro6o/p3MbKwf166UHMDk/GuV5Djimk3ruB8Z3p41mZtZzetOYjpmZ9XEOOmZmVhkHHTMzq4yDjpmZVcZBx8zMKuOgY2ZmlXHQMTOzyjjomJlZZRx0zMysMg46ZmZWGQcdMzOrjIOOmZlVxkHHzMwq46BjZmaVcdAxM7PKOOiYmVllHHTMzKwyDjpmZlYZBx0zM6uMg46ZmVXGQcfMzCrjoGNmZpVx0DEzs8o46JiZWWUcdMzMrDIOOmZmVhkHHTMzq4yDjpmZVaZXBR1JkyVF6TGnkK6cZ7akpZJuk7RzqYzNJV0uaWF+XC5ps+r3xszMynpV0MkeAbYuPHYppJ0KfA44CdgTmAdMlbRJIc+VwB7AocAh+f/Le77ZZmbWmXXb3YA6Xo2IOeWVkgR8FvhORFyd1x1HCjxHARdJGkUKNPtFxLSc5+PAHZJ2jIhHqtoJMzNbXW/s6WwvaZakxyX9XNL2ef12wDDg5lrGiFgK3A7sk1eNAxYB0wrl3QksLuQxM7M26W1B50/AJNKtsY+Rgsw0SW/I/wPMLW0zt5A2DJgfEVFLzP/PK+RZhaQTJE2XNH3+/Pmt2g8zM6ujV91ei4jfFZcl3Q08BhwH3F3LVtpMpXXl9Hp5inVeDFwMMHbs2Lp5zMysNXpbT2cVEbEIeBB4M1Ab5yn3WIaysvczBxiax3+A18eCtmT1HpKZmVWsVwcdSesDbwGeAR4nBZUJpfT9WTmGcxewMWlsp2YcsBGrjvOYmVkb9Krba5LOBK4FniL1YP4fKWBcGhEh6WzgS5L+Bvwd+DJp4sCVABHxsKQbSTPZPka6rXYRcJ1nrpmZtV+vCjrAG4GfAUOA+aRxnL0j4smcfjpUl+HAAAAJEElEQVSwAXA+sDlp4sHEiHipUMbRwLmsnOV2DfCpnm+6mZl1plcFnYj4QCfpAUzOj0Z5ngOOaWnDzMysJXr1mI6ZmfUtDjpmZlYZBx0zM6uMg46ZmVXGQcfMzCrjoGNmZpVx0DEzs8o46JiZWWUcdMzMrDIOOmZmVhkHHTMzq4yDjpmZVcZBx8zMKuOgY2ZmlXHQMTOzyjjomJlZZRx0zMysMg46ZmZWGQcdMzOrjIOOmZlVZt12N8DMrM+bvGnF9S2str4ucE/HzMwq46BjZmaVcdAxM7PKOOiYmVllHHTMzKwyDjpmZlaZPh10JJ0o6XFJL0uaIWn/drfJzKw/67NBR9L7gXOAbwG7A9OA30ka0daGmZn1Y335w6EnA1Mi4pK8fJKkQ4BPAl9oX7PMrDcY+fnrK6vrifUrq6rX65M9HUmDgDHAzaWkm4F9qm+RmZkBKCLa3YaWk7QNMAt4e0TcXlj/FeDoiNixsO4E4IS8uCPwSJVtrdgQYEG7G2Hd5udv7dXXn7ttI2LLZjL25dtrAOWIqvK6iLgYuLiyFrWRpOkRMbbd7bDu8fO39vJzt1KfvL1GekexAhhWWj8UmFt9c8zMDPpo0ImIV4AZwIRS0gTSLDYzM2uDvnx77Szgckn/B9wJfALYBriwra1qr35xG7EP8/O39vJzl/XJiQQ1kk4ETgW2Bh4A/qs4scDMzKrVp4OOmZn1Ln1yTMfMzHonBx0zM6uMg04fJWnd/IWn27S7LWZmNR7T6cMkLQZ2iogn290WMzPo21OmDe4G9gAcdNZSkt4GvIP0weZV7kxExKfb0ijrMkkbA0TEona3pd0cdPq2S4DvSdqW9GHZxcXEiLinLa2ypkg6BTgdeBSYzapf4eRbFGsBSZ8lfeP98Lw8m/QZwrOjn95m8u21PkzSax0kR0QMqKwx1mWSnga+GxE/aHdbrOsknU76MuEzgLvy6nHAKcAlEXFqu9rWTg46fVju4TTksZ7eTdJCYPeIeKzdbbGuk/QccEJE/LK0/n3ARRHxhva0rL18e60Pi4gnJa0L7AWMAAYVk4HL29Iwa9bPgEOAH7a7IdZt9zVY129nDrun04dJegtwLbAd6WcdVpDeaCwHlkXE4DY2z+qQdHJhcQPgs6QfH7yP9Ly9LiLOqrBp1kWSziZdYz9TWv99YEB/nQjioNOHSboReAH4CDAH2A3YFLgA+HJETG1j86wOSY83mTUiYvsebYytEUkXAEcBz5BmkgK8jfTFw1cAr9by9qcA5KDTh0l6lvTrqQ/k8YG9IuIRSW8HzouIXdvcRLM+S9KtTWaNiDioRxvTi3hMp28TsCT/P580bfMRYCawQ7saZc3JP69+ZkQsKa3fAPjviPhae1pmzYiIA9vdht7IPZ0+TNLtwPcj4teSrgTeAHwL+Biwq3s6vZukFcDWETGvtP4NwDxPee99JF0DHBMRL+b/G4mIOKKqdvUm7un0bd8ENsr/fxm4DriV9HPe/9GuRlnTRP0Pge4OPFdxW6w5z7LyOXu2nQ3prdzT6WckbQE8318/Db02kPQS6cK1Een2aPG5GgCsD1wYEf/ZhuaZrREHHbNeRtJxpF7OT0hTphcWkl8BnoiIu+pta9bbOeiY9VJ5luG0iFjeaWaztYSDjlkvJmk94GhgJ9JttgeBn0XEsrY2zKybHHTMeilJOwE3AoOB+/PqXUi32w6JiIfb1Taz7nLQMeulJE0lTSQ4NiJezOsGAz8F1ouIg9vZPrPucNAx66UkLQH2jIgHS+t3Ae6OiI3qb2nWe/Xbbzo1Wwu8DGxWZ/2mOc1sreOgY9Z7XQtcImlfSQPyYz/gIqCjT7ub9Vq+vWbWS0naDLgUeBfpZykgfTj0t8CkiFjYaFuz3spBx6yXk7QDMCovPhQR/2xne8zWhIOOWS8m6bPAyaRvCAeYDZwFnO2vMrK1kb/w06yXknQ6cAJwBlD72ptxwFeArYFT29Q0s25zT8esl5L0HHBCRPyytP59wEUR8Yb2tMys+zx7zax3u6/BOr92ba3kE9es97oMqPfzBZ8ELq+4LWYt4TEds15E0rmFxXWBYyQdDNyd170N2Aa4ouq2mbWCx3TMehFJtzaZNSLioB5tjFkPcNAxM7PKeEzHzMwq46BjZmaVcdAxq4ik6yRNaWF5T0g6pVXlmVXBs9fM1l57Aovb3QizrnDQMVtLRcT8drfBrKt8e82sB0jaUNIUSYskzZX0xVL6IEnflTRT0mJJf86fx0HSOnn9SaVt/lVSSNo9L69ye03SYEkXSHpG0suSHpb0/kL6PpL+KGmJpFk57+CePRJmq3LQMesZZwITgPcC7wB2B8YX0v8HeDtwFLAL6XdzrpX01oh4DfgZcHSpzKNJP23wl3JlkgT8Lpf5YWAn0rdTv5LTdwFuJv3421uBI4HdgJ+0YF/NmubP6Zi1mKSNgWeB4yPiisK6mcBvgK8D/wBGRsRThe1+A8yOiBMl7Qr8FXhzRDya0/8B/CQivp2XnwB+EBFnSpoA3ATsHBEP12nTZcDyiPhIYd1uwF+ArSJiXquPg1k97umYtd6bgEGs/DkCImIRcH9e3AMQ8FC+/bZI0iLg8LwtEXFfzn8UgKS35bQrG9S5O/BMvYCTjSF9pU6xvjsL7TWrhCcSmLWeOklfBwjS7LPlpbSlhf+vAI4Hvka6tXZHRDy5BnX+CPh+nbRZnWxr1jIOOmat9ygpmOwNPAYgaSNgNPBP0i0tAcMioqPvWrsC+JakvYH3A1/uIO89wNaSRjXo7dxDuvX2aFd3xqyVfHvNrMXyrbQfA9+VNEHSzqQB+wE5/e+kgDJF0vskbS9prKRTJB1ZKGcmcDtwIbAp8IsOqv098CfgakkHS9ou1/2enP5dYC9JF0raXdIOkt4p6aIW775Zhxx0zHrGKcCtwK/z3wdIAaTmw6QZbKcDfwOuI81uK98+u5w02+z6iHihUWV5xtuhpHGanwIPA+eQxpZqY0TjgZHAH0mTFL4NzO3+Lpp1nWevmZlZZdzTMTOzyjjomJlZZRx0zMysMg46ZmZWGQcdMzOrjIOOmZlVxkHHzMwq46BjZmaVcdAxM7PK/H+ms47RjK39uwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x3363b90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure()\n",
    "ax = fig.add_subplot(111)\n",
    "dc = device_feature.groupby(['device', 'Churn'])['uid'].count().unstack()\n",
    "print(dc)\n",
    "dcsum = dc.sum()\n",
    "dc.plot(kind = 'bar', ax= ax)\n",
    "plt.title('Distribution of Churn/Active for 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['uid']+1, v['uid'],ha='left', va='bottom', fontsize = 12)\n",
    "#    ax.text(i,(v['uid']/dcsum).round(3) + 1, (v['uid']/dcsum).round(3),ha='center',fontsize = 12)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.3. merge device_feature and play_feature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "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>device</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>uid</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>168651712</th>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154519904</th>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168453408</th>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>125549120</th>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167621616</th>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168676064</th>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>155812384</th>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>155769904</th>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          device\n",
       "uid             \n",
       "168651712     ar\n",
       "154519904   both\n",
       "168453408   both\n",
       "125549120     ar\n",
       "167621616     ar\n",
       "168676064   both\n",
       "155812384     ar\n",
       "155769904     ar"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "uid_device = sample_play[['uid','device']].drop_duplicates()\n",
    "uid_device.set_index('uid', inplace = True)\n",
    "uid_device.shape\n",
    "uid_device.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(8798, 14)\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>freq_play_last_30_days</th>\n",
       "      <th>avg_playtime_last_30_days</th>\n",
       "      <th>freq_play_last_21_days</th>\n",
       "      <th>avg_playtime_last_21_days</th>\n",
       "      <th>freq_play_last_14_days</th>\n",
       "      <th>avg_playtime_last_14_days</th>\n",
       "      <th>freq_play_last_7_days</th>\n",
       "      <th>avg_playtime_last_7_days</th>\n",
       "      <th>freq_play_last_3_days</th>\n",
       "      <th>avg_playtime_last_3_days</th>\n",
       "      <th>freq_play_last_1_days</th>\n",
       "      <th>avg_playtime_last_1_days</th>\n",
       "      <th>Churn</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",
       "      <th></th>\n",
       "      <th></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>168362736</th>\n",
       "      <td>1568.0</td>\n",
       "      <td>152.141808</td>\n",
       "      <td>1071.0</td>\n",
       "      <td>157.697479</td>\n",
       "      <td>810.0</td>\n",
       "      <td>161.491358</td>\n",
       "      <td>379.0</td>\n",
       "      <td>164.002639</td>\n",
       "      <td>113.0</td>\n",
       "      <td>168.433628</td>\n",
       "      <td>68.0</td>\n",
       "      <td>197.75</td>\n",
       "      <td>0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167706656</th>\n",
       "      <td>461.0</td>\n",
       "      <td>158.316703</td>\n",
       "      <td>237.0</td>\n",
       "      <td>143.274262</td>\n",
       "      <td>149.0</td>\n",
       "      <td>143.798658</td>\n",
       "      <td>51.0</td>\n",
       "      <td>71.647059</td>\n",
       "      <td>26.0</td>\n",
       "      <td>59.653846</td>\n",
       "      <td>8.0</td>\n",
       "      <td>99.50</td>\n",
       "      <td>0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154788992</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154794176</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154657184</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168559184</th>\n",
       "      <td>358.0</td>\n",
       "      <td>191.466480</td>\n",
       "      <td>82.0</td>\n",
       "      <td>178.682927</td>\n",
       "      <td>9.0</td>\n",
       "      <td>135.888889</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168036928</th>\n",
       "      <td>1511.0</td>\n",
       "      <td>174.262078</td>\n",
       "      <td>727.0</td>\n",
       "      <td>188.653370</td>\n",
       "      <td>438.0</td>\n",
       "      <td>189.029680</td>\n",
       "      <td>271.0</td>\n",
       "      <td>194.317343</td>\n",
       "      <td>66.0</td>\n",
       "      <td>156.515152</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154745184</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           freq_play_last_30_days  avg_playtime_last_30_days  \\\n",
       "uid                                                            \n",
       "168362736                  1568.0                 152.141808   \n",
       "167706656                   461.0                 158.316703   \n",
       "154788992                     NaN                        NaN   \n",
       "154794176                     NaN                        NaN   \n",
       "154657184                     NaN                        NaN   \n",
       "168559184                   358.0                 191.466480   \n",
       "168036928                  1511.0                 174.262078   \n",
       "154745184                     NaN                        NaN   \n",
       "\n",
       "           freq_play_last_21_days  avg_playtime_last_21_days  \\\n",
       "uid                                                            \n",
       "168362736                  1071.0                 157.697479   \n",
       "167706656                   237.0                 143.274262   \n",
       "154788992                     NaN                        NaN   \n",
       "154794176                     NaN                        NaN   \n",
       "154657184                     NaN                        NaN   \n",
       "168559184                    82.0                 178.682927   \n",
       "168036928                   727.0                 188.653370   \n",
       "154745184                     NaN                        NaN   \n",
       "\n",
       "           freq_play_last_14_days  avg_playtime_last_14_days  \\\n",
       "uid                                                            \n",
       "168362736                   810.0                 161.491358   \n",
       "167706656                   149.0                 143.798658   \n",
       "154788992                     NaN                        NaN   \n",
       "154794176                     NaN                        NaN   \n",
       "154657184                     NaN                        NaN   \n",
       "168559184                     9.0                 135.888889   \n",
       "168036928                   438.0                 189.029680   \n",
       "154745184                     NaN                        NaN   \n",
       "\n",
       "           freq_play_last_7_days  avg_playtime_last_7_days  \\\n",
       "uid                                                          \n",
       "168362736                  379.0                164.002639   \n",
       "167706656                   51.0                 71.647059   \n",
       "154788992                    NaN                       NaN   \n",
       "154794176                    NaN                       NaN   \n",
       "154657184                    NaN                       NaN   \n",
       "168559184                    NaN                       NaN   \n",
       "168036928                  271.0                194.317343   \n",
       "154745184                    NaN                       NaN   \n",
       "\n",
       "           freq_play_last_3_days  avg_playtime_last_3_days  \\\n",
       "uid                                                          \n",
       "168362736                  113.0                168.433628   \n",
       "167706656                   26.0                 59.653846   \n",
       "154788992                    NaN                       NaN   \n",
       "154794176                    NaN                       NaN   \n",
       "154657184                    NaN                       NaN   \n",
       "168559184                    NaN                       NaN   \n",
       "168036928                   66.0                156.515152   \n",
       "154745184                    NaN                       NaN   \n",
       "\n",
       "           freq_play_last_1_days  avg_playtime_last_1_days  Churn device  \n",
       "uid                                                                       \n",
       "168362736                   68.0                    197.75      0   both  \n",
       "167706656                    8.0                     99.50      0   both  \n",
       "154788992                    NaN                       NaN      1   both  \n",
       "154794176                    NaN                       NaN      1   both  \n",
       "154657184                    NaN                       NaN      1   both  \n",
       "168559184                    NaN                       NaN      1   both  \n",
       "168036928                    NaN                       NaN      0     ar  \n",
       "154745184                    NaN                       NaN      1   both  "
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "play_feature_all = play_feature_8798.join(uid_device,how = \"left\")\n",
    "print(play_feature_all.shape)\n",
    "play_feature_all.sample(8)\n"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "# why the number of uid drop after generating freq feature??\n",
    "# Because there are churn users never show up during this period, thus, they are filtered, that's why the number of churn drops largely, thus to keep those churn user still in, after generate all features, add those user back to all_feature, and fillna(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "How many active user after feature generation: 4397\n",
      "How many churn user after feature generation: 891\n"
     ]
    }
   ],
   "source": [
    "#################Original#################\n",
    "print('How many active user after feature generation:',len(play_feature_all[play_feature_all.Churn == 0]))\n",
    "print('How many churn user after feature generation:',len(play_feature_all[play_feature_all.Churn == 1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "How many active user after feature generation: 4465\n",
      "How many churn user after feature generation: 4358\n"
     ]
    }
   ],
   "source": [
    "###################Adjusted##################\n",
    "print('How many active user after feature generation:',len(play_feature_all[play_feature_all.Churn == 0]))\n",
    "print('How many churn user after feature generation:',len(play_feature_all[play_feature_all.Churn == 1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8798, 12)"
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#double check the play_feature\n",
    "play_feature_all.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.4 complete_degree\n",
    "check average music play complete degree of each user: avg(play_time/song_length)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 225,
   "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>play_time</th>\n",
       "      <th>device</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1340831</th>\n",
       "      <td>167790352</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2017-04-11</td>\n",
       "      <td>0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1882433</th>\n",
       "      <td>168576592</td>\n",
       "      <td>11819020.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>2017-03-30</td>\n",
       "      <td>1</td>\n",
       "      <td>12.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1028167</th>\n",
       "      <td>154730000</td>\n",
       "      <td>2696023.0</td>\n",
       "      <td>208.0</td>\n",
       "      <td>2017-03-01</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3959728</th>\n",
       "      <td>154775840</td>\n",
       "      <td>6227698.0</td>\n",
       "      <td>69.0</td>\n",
       "      <td>2017-03-06</td>\n",
       "      <td>1</td>\n",
       "      <td>11.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>131436</th>\n",
       "      <td>168684560</td>\n",
       "      <td>6060869.0</td>\n",
       "      <td>391.0</td>\n",
       "      <td>2017-04-12</td>\n",
       "      <td>1</td>\n",
       "      <td>110.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1423014</th>\n",
       "      <td>168874112</td>\n",
       "      <td>157908.0</td>\n",
       "      <td>306.0</td>\n",
       "      <td>2017-04-27</td>\n",
       "      <td>0</td>\n",
       "      <td>306.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3692807</th>\n",
       "      <td>168809664</td>\n",
       "      <td>6807407.0</td>\n",
       "      <td>215.0</td>\n",
       "      <td>2017-03-30</td>\n",
       "      <td>0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>ar</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2330882</th>\n",
       "      <td>167854176</td>\n",
       "      <td>504089.0</td>\n",
       "      <td>91.0</td>\n",
       "      <td>2017-04-03</td>\n",
       "      <td>0</td>\n",
       "      <td>91.0</td>\n",
       "      <td>both</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               uid     song_id  song_length       date  Churn  play_time  \\\n",
       "1340831  167790352         0.0          4.0 2017-04-11      0        4.0   \n",
       "1882433  168576592  11819020.0         12.0 2017-03-30      1       12.0   \n",
       "1028167  154730000   2696023.0        208.0 2017-03-01      1        1.0   \n",
       "3959728  154775840   6227698.0         69.0 2017-03-06      1       11.0   \n",
       "131436   168684560   6060869.0        391.0 2017-04-12      1      110.0   \n",
       "1423014  168874112    157908.0        306.0 2017-04-27      0      306.0   \n",
       "3692807  168809664   6807407.0        215.0 2017-03-30      0        5.0   \n",
       "2330882  167854176    504089.0         91.0 2017-04-03      0       91.0   \n",
       "\n",
       "        device  \n",
       "1340831     ar  \n",
       "1882433   both  \n",
       "1028167   both  \n",
       "3959728     ar  \n",
       "131436    both  \n",
       "1423014   both  \n",
       "3692807     ar  \n",
       "2330882   both  "
      ]
     },
     "execution_count": 225,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_play.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 226,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "How many user play time =0: play_time      0\n",
      "song_length    0\n",
      "dtype: int64\n",
      "          play_time  song_length\n",
      "uid                             \n",
      "838955   181.750000   233.833333\n",
      "971787    32.500000   243.000000\n",
      "1264383   87.666667   234.666667\n",
      "1901839  167.249548   197.471971\n",
      "2163514  159.500000   248.824324\n"
     ]
    }
   ],
   "source": [
    "music_complete = sample_play.groupby('uid')[['play_time', 'song_length']].mean()\n",
    "print('How many user play time =0:', music_complete[music_complete.play_time ==0].count())\n",
    "print(music_complete.head())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 227,
   "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>precent</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>838955</td>\n",
       "      <td>0.777263</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>971787</td>\n",
       "      <td>0.133745</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1264383</td>\n",
       "      <td>0.373580</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1901839</td>\n",
       "      <td>0.846953</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2163514</td>\n",
       "      <td>0.641015</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       uid   precent\n",
       "0   838955  0.777263\n",
       "1   971787  0.133745\n",
       "2  1264383  0.373580\n",
       "3  1901839  0.846953\n",
       "4  2163514  0.641015"
      ]
     },
     "execution_count": 227,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "complete_precent = music_complete.play_time/music_complete.song_length\n",
    "complete_precent = pd.DataFrame(complete_precent, columns = ['precent'])\n",
    "complete_precent.reset_index(inplace = True)\n",
    "complete_precent.head()\n",
    "#there are 25 user with play_time =0, 928 user with play/song_length >1."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Series([], Name: precent, dtype: float64)"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#check whether have out of [0,1]\n",
    "complete_precent.precent[(complete_precent.precent==0)|(complete_precent.precent>1)]\n",
    "#OK"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The play_complete mean is 0.6319226954988768\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Text(0,0.5,'Count')"
      ]
     },
     "execution_count": 106,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x27648670>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plot precent\n",
    "complete_precent['precent'].hist(bins =30, alpha = 0.9)\n",
    "plt.title('Histogram of play_complete(%)' , fontsize = 13)\n",
    "print(\"The play_complete mean is\", complete_precent['precent'].mean())\n",
    "plt.axvline(complete_precent['precent'].mean(), color='blue',alpha = 0.9, linestyle='dashed', linewidth=2)\n",
    "plt.xlabel(\"play_complete(%)\", fontsize = 13)\n",
    "plt.ylabel(\"Count\", fontsize = 13)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.4. save play_feature_all"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "#this is 8823, remember to fillna with 0\n",
    "play_feature_all.to_csv(\"D:/MusicFile/play_feature_8798_30.csv\", header = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. for search and download.log\n",
    "### 2.1 frequency feature generation for search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "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>file_name</th>\n",
       "      <th>date</th>\n",
       "      <th>record_time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>248324</th>\n",
       "      <td>133928864.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>20170430_3_search.log</td>\n",
       "      <td>2017-04-30T00:00:00.000-04:00</td>\n",
       "      <td>2017-04-30T21:05:30.000-04:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>438828</th>\n",
       "      <td>167849488.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>20170330_1_search.log</td>\n",
       "      <td>2017-03-30T00:00:00.000-04:00</td>\n",
       "      <td>2017-03-30T18:31:34.000-04:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25669</th>\n",
       "      <td>168668896.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>20170417_3_search.log</td>\n",
       "      <td>2017-04-17T00:00:00.000-04:00</td>\n",
       "      <td>2017-04-17T13:49:22.000-04:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92134</th>\n",
       "      <td>168174560.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>20170412_2_search.log</td>\n",
       "      <td>2017-04-12T00:00:00.000-04:00</td>\n",
       "      <td>2017-04-12T13:27:56.000-04:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>313678</th>\n",
       "      <td>168880640.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>20170330_3_search.log</td>\n",
       "      <td>2017-03-30T00:00:00.000-04:00</td>\n",
       "      <td>2017-03-30T20:01:05.000-04:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>74365</th>\n",
       "      <td>154630352.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>1_4_search.log</td>\n",
       "      <td>2017-03-04T00:00:00.000-05:00</td>\n",
       "      <td>2017-03-04T14:11:18.000-05:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115616</th>\n",
       "      <td>168307328.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>20170413_2_search.log</td>\n",
       "      <td>2017-04-13T00:00:00.000-04:00</td>\n",
       "      <td>2017-04-13T15:25:52.000-04:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>135949</th>\n",
       "      <td>167642272.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>20170401_1_search.log</td>\n",
       "      <td>2017-04-01T00:00:00.000-04:00</td>\n",
       "      <td>2017-04-01T17:47:05.000-04:00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                uid device              file_name  \\\n",
       "248324  133928864.0     ar  20170430_3_search.log   \n",
       "438828  167849488.0     ar  20170330_1_search.log   \n",
       "25669   168668896.0     ar  20170417_3_search.log   \n",
       "92134   168174560.0     ar  20170412_2_search.log   \n",
       "313678  168880640.0     ar  20170330_3_search.log   \n",
       "74365   154630352.0     ar         1_4_search.log   \n",
       "115616  168307328.0     ar  20170413_2_search.log   \n",
       "135949  167642272.0     ar  20170401_1_search.log   \n",
       "\n",
       "                                 date                    record_time  \n",
       "248324  2017-04-30T00:00:00.000-04:00  2017-04-30T21:05:30.000-04:00  \n",
       "438828  2017-03-30T00:00:00.000-04:00  2017-03-30T18:31:34.000-04:00  \n",
       "25669   2017-04-17T00:00:00.000-04:00  2017-04-17T13:49:22.000-04:00  \n",
       "92134   2017-04-12T00:00:00.000-04:00  2017-04-12T13:27:56.000-04:00  \n",
       "313678  2017-03-30T00:00:00.000-04:00  2017-03-30T20:01:05.000-04:00  \n",
       "74365   2017-03-04T00:00:00.000-05:00  2017-03-04T14:11:18.000-05:00  \n",
       "115616  2017-04-13T00:00:00.000-04:00  2017-04-13T15:25:52.000-04:00  \n",
       "135949  2017-04-01T00:00:00.000-04:00  2017-04-01T17:47:05.000-04:00  "
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sampled_search = pd.read_csv(\"D:/MusicFile/sample_search_316.csv\")\n",
    "sampled_search.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [],
   "source": [
    "sampled_search['date'] = pd.to_datetime(sampled_search['date'].str.slice(0,10))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "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>file_name</th>\n",
       "      <th>date</th>\n",
       "      <th>record_time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>174544</th>\n",
       "      <td>168032048.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>20170405_1_search.log</td>\n",
       "      <td>2017-04-05</td>\n",
       "      <td>2017-04-05T10:43:39.000-04:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>331657</th>\n",
       "      <td>154564032.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>1_1_search.log</td>\n",
       "      <td>2017-03-01</td>\n",
       "      <td>2017-03-01T16:10:14.000-05:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>93294</th>\n",
       "      <td>154794272.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>1_1_search.log</td>\n",
       "      <td>2017-03-01</td>\n",
       "      <td>2017-03-01T23:29:15.000-05:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>468830</th>\n",
       "      <td>168879056.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>20170330_3_search.log</td>\n",
       "      <td>2017-03-30</td>\n",
       "      <td>2017-03-30T20:14:16.000-04:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>337112</th>\n",
       "      <td>169040624.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>20170419_3_search.log</td>\n",
       "      <td>2017-04-19</td>\n",
       "      <td>2017-04-19T09:31:00.000-04:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>367342</th>\n",
       "      <td>154662560.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>1_1_search.log</td>\n",
       "      <td>2017-03-01</td>\n",
       "      <td>2017-03-01T19:13:47.000-05:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99225</th>\n",
       "      <td>168376224.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>20170420_2_search.log</td>\n",
       "      <td>2017-04-20</td>\n",
       "      <td>2017-04-20T18:24:33.000-04:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>213695</th>\n",
       "      <td>167907184.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>20170416_1_search.log</td>\n",
       "      <td>2017-04-16</td>\n",
       "      <td>2017-04-16T23:44:42.000-04:00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                uid device              file_name       date  \\\n",
       "174544  168032048.0     ar  20170405_1_search.log 2017-04-05   \n",
       "331657  154564032.0     ar         1_1_search.log 2017-03-01   \n",
       "93294   154794272.0     ar         1_1_search.log 2017-03-01   \n",
       "468830  168879056.0     ar  20170330_3_search.log 2017-03-30   \n",
       "337112  169040624.0     ar  20170419_3_search.log 2017-04-19   \n",
       "367342  154662560.0     ar         1_1_search.log 2017-03-01   \n",
       "99225   168376224.0     ar  20170420_2_search.log 2017-04-20   \n",
       "213695  167907184.0     ar  20170416_1_search.log 2017-04-16   \n",
       "\n",
       "                          record_time  \n",
       "174544  2017-04-05T10:43:39.000-04:00  \n",
       "331657  2017-03-01T16:10:14.000-05:00  \n",
       "93294   2017-03-01T23:29:15.000-05:00  \n",
       "468830  2017-03-30T20:14:16.000-04:00  \n",
       "337112  2017-04-19T09:31:00.000-04:00  \n",
       "367342  2017-03-01T19:13:47.000-05:00  \n",
       "99225   2017-04-20T18:24:33.000-04:00  \n",
       "213695  2017-04-16T23:44:42.000-04:00  "
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sampled_search.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "uid                   float64\n",
       "device                 object\n",
       "file_name              object\n",
       "date           datetime64[ns]\n",
       "record_time            object\n",
       "dtype: object"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sampled_search.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7714"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sampled_search = sampled_search[['uid', 'device', 'date']]\n",
    "len(sampled_search.uid.unique())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [],
   "source": [
    "sampled_search.to_csv(\"D:/MusicFile/sampled_search_316new.csv\", header = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "'''\n",
    "def get_freq_avg(df, log_type, window, snapshot_date):\n",
    "    snapshot_date = pd.to_datetime(snapshot_date)\n",
    "    df2 = df.loc[(snapshot_date - df['date'])<= dt.timedelta(window), :]\n",
    "    df_feature = df2.groupby('uid')['play_time'] \\\n",
    "    .agg(['count', 'mean']) \\\n",
    "    .rename(columns = {'count':'freq_' + log_type + '_last_' + str(window) + '_days',\n",
    "                       'mean' : 'avg_'+ log_type + 'time' + '_last_' + str(window) + '_days'})\n",
    "    return df_feature\n",
    "\n",
    "def get_all_freq(df, log_type, window_list, snapshot_time):\n",
    "    type_feature = get_freq_avg(df, log_type, window_list[0], snapshot_date)\n",
    "    for window in window_list[1:]:\n",
    "        test_what = get_freq_avg(df, log_type, window, snapshot_date\n",
    "        type_feature = pd.merge(test_what, type_feature, how='left', left_index=True, right_index=True)\n",
    "    return type_feature\n",
    "\n",
    "\n",
    "log_types = ['play', 'search', 'download']\n",
    "window_list = [1,3,7,14,21]\n",
    "snapshot_date = '2017-04-29'\n",
    "\n",
    "'''\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "log_types = ['play', 'search', 'download']\n",
    "window_list = [1,3,7,14,21,30]\n",
    "snapshot_date = '2017-04-29'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_freq(df, log_type, window, snapshot_time):\n",
    "    snapshot_date = pd.to_datetime(snapshot_time)\n",
    "    df2 = df.loc[(snapshot_date - df['date'])<= dt.timedelta(window), :]\n",
    "    df_feature = df2.groupby('uid')['date'] \\\n",
    "                    .agg(['count']) \\\n",
    "                    .rename(columns = {'count':'freq_' + log_type + '_last_' + str(window) + '_days'})\n",
    "    \n",
    "    return df_feature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "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>freq_search_last_1_days</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>uid</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1901839.0</th>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2163514.0</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2285308.0</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3752046.0</th>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5597959.0</th>\n",
       "      <td>18</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           freq_search_last_1_days\n",
       "uid                               \n",
       "1901839.0                        5\n",
       "2163514.0                        1\n",
       "2285308.0                        1\n",
       "3752046.0                        3\n",
       "5597959.0                       18"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sest = get_freq(sampled_search, 'search', window_list[0], snapshot_date)\n",
    "sest.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_all_freq(df, log_type, window_list, snapshot_time):\n",
    "    type_feature = get_freq(df, log_type, window_list[0], snapshot_date)\n",
    "    for window in window_list[1:]:\n",
    "        test_what = get_freq(df, log_type, window, snapshot_date)\n",
    "        type_feature = pd.merge(test_what, type_feature, how='left', left_index=True, right_index=True)\n",
    "    return type_feature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "search_feature = get_all_freq(sampled_search, 'search', window_list, snapshot_date)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "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>freq_search_last_30_days</th>\n",
       "      <th>freq_search_last_21_days</th>\n",
       "      <th>freq_search_last_14_days</th>\n",
       "      <th>freq_search_last_7_days</th>\n",
       "      <th>freq_search_last_3_days</th>\n",
       "      <th>freq_search_last_1_days</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>151486256.0</th>\n",
       "      <td>3</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168048112.0</th>\n",
       "      <td>11</td>\n",
       "      <td>10.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168489712.0</th>\n",
       "      <td>110</td>\n",
       "      <td>18.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>8.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>139851616.0</th>\n",
       "      <td>114</td>\n",
       "      <td>35.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168577888.0</th>\n",
       "      <td>3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168985984.0</th>\n",
       "      <td>33</td>\n",
       "      <td>12.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168757824.0</th>\n",
       "      <td>312</td>\n",
       "      <td>179.0</td>\n",
       "      <td>164.0</td>\n",
       "      <td>126.0</td>\n",
       "      <td>117.0</td>\n",
       "      <td>115.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168319296.0</th>\n",
       "      <td>79</td>\n",
       "      <td>64.0</td>\n",
       "      <td>56.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>41.0</td>\n",
       "      <td>34.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             freq_search_last_30_days  freq_search_last_21_days  \\\n",
       "uid                                                               \n",
       "151486256.0                         3                       3.0   \n",
       "168048112.0                        11                      10.0   \n",
       "168489712.0                       110                      18.0   \n",
       "139851616.0                       114                      35.0   \n",
       "168577888.0                         3                       NaN   \n",
       "168985984.0                        33                      12.0   \n",
       "168757824.0                       312                     179.0   \n",
       "168319296.0                        79                      64.0   \n",
       "\n",
       "             freq_search_last_14_days  freq_search_last_7_days  \\\n",
       "uid                                                              \n",
       "151486256.0                       1.0                      1.0   \n",
       "168048112.0                      10.0                     10.0   \n",
       "168489712.0                      13.0                     11.0   \n",
       "139851616.0                       1.0                      NaN   \n",
       "168577888.0                       NaN                      NaN   \n",
       "168985984.0                      12.0                      7.0   \n",
       "168757824.0                     164.0                    126.0   \n",
       "168319296.0                      56.0                     42.0   \n",
       "\n",
       "             freq_search_last_3_days  freq_search_last_1_days  \n",
       "uid                                                            \n",
       "151486256.0                      1.0                      1.0  \n",
       "168048112.0                     10.0                      1.0  \n",
       "168489712.0                      8.0                      8.0  \n",
       "139851616.0                      NaN                      NaN  \n",
       "168577888.0                      NaN                      NaN  \n",
       "168985984.0                      5.0                      4.0  \n",
       "168757824.0                    117.0                    115.0  \n",
       "168319296.0                     41.0                     34.0  "
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "search_feature.sample(8)\n",
    "#notice: uid now is the row index, we need it as a column uid, but when save to csv, it automatically change to a column"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5211, 6)"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#same as  play_feature, there are some churn user never show up during the window time\n",
    "search_feature.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "search_feature.to_csv(\"D:/MusicFile/search_feature_318.csv\", header = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.2 frequency feature generation for download"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "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>date</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>193736</th>\n",
       "      <td>168877040.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>23532356.0</td>\n",
       "      <td>2017-04-06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>84347</th>\n",
       "      <td>168703680.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>5185874.0</td>\n",
       "      <td>2017-05-10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9652</th>\n",
       "      <td>168287040.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>708287.0</td>\n",
       "      <td>2017-04-02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>166277</th>\n",
       "      <td>169038144.0</td>\n",
       "      <td>ip</td>\n",
       "      <td>19087234.0</td>\n",
       "      <td>2017-03-30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>76435</th>\n",
       "      <td>39560840.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>6585112.0</td>\n",
       "      <td>2017-04-03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>277633</th>\n",
       "      <td>167906032.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>6657692.0</td>\n",
       "      <td>2017-04-15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>162043</th>\n",
       "      <td>168574288.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>13800660.0</td>\n",
       "      <td>2017-03-30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>247365</th>\n",
       "      <td>168490624.0</td>\n",
       "      <td>ar</td>\n",
       "      <td>176292.0</td>\n",
       "      <td>2017-04-03</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                uid device     song_id       date\n",
       "193736  168877040.0     ar  23532356.0 2017-04-06\n",
       "84347   168703680.0     ar   5185874.0 2017-05-10\n",
       "9652    168287040.0     ar    708287.0 2017-04-02\n",
       "166277  169038144.0     ip  19087234.0 2017-03-30\n",
       "76435    39560840.0     ar   6585112.0 2017-04-03\n",
       "277633  167906032.0     ar   6657692.0 2017-04-15\n",
       "162043  168574288.0     ar  13800660.0 2017-03-30\n",
       "247365  168490624.0     ar    176292.0 2017-04-03"
      ]
     },
     "execution_count": 96,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sampled_download = pd.read_csv(\"D:/MusicFile/sample_download_316.csv\")\n",
    "sampled_download['date'] = pd.to_datetime(sampled_download['date'].str.slice(0,10))\n",
    "sampled_download.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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>freq_download_last_1_days</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>uid</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2280694.0</th>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3752046.0</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7910960.0</th>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8853855.0</th>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14780539.0</th>\n",
       "      <td>27</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            freq_download_last_1_days\n",
       "uid                                  \n",
       "2280694.0                           5\n",
       "3752046.0                           1\n",
       "7910960.0                           3\n",
       "8853855.0                           2\n",
       "14780539.0                         27"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dest = get_freq(sampled_download, 'download', window_list[0], snapshot_date)\n",
    "dest.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "download_feature = get_all_freq(sampled_download, 'download', window_list, snapshot_date)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "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>freq_download_last_30_days</th>\n",
       "      <th>freq_download_last_21_days</th>\n",
       "      <th>freq_download_last_14_days</th>\n",
       "      <th>freq_download_last_7_days</th>\n",
       "      <th>freq_download_last_3_days</th>\n",
       "      <th>freq_download_last_1_days</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>167905248.0</th>\n",
       "      <td>13</td>\n",
       "      <td>7.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168302176.0</th>\n",
       "      <td>55</td>\n",
       "      <td>40.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>169027984.0</th>\n",
       "      <td>527</td>\n",
       "      <td>286.0</td>\n",
       "      <td>165.0</td>\n",
       "      <td>74.0</td>\n",
       "      <td>64.0</td>\n",
       "      <td>60.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168110000.0</th>\n",
       "      <td>4</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>140533488.0</th>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168434896.0</th>\n",
       "      <td>15</td>\n",
       "      <td>13.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>151637920.0</th>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168509952.0</th>\n",
       "      <td>12</td>\n",
       "      <td>12.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>12.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             freq_download_last_30_days  freq_download_last_21_days  \\\n",
       "uid                                                                   \n",
       "167905248.0                          13                         7.0   \n",
       "168302176.0                          55                        40.0   \n",
       "169027984.0                         527                       286.0   \n",
       "168110000.0                           4                         2.0   \n",
       "140533488.0                           1                         NaN   \n",
       "168434896.0                          15                        13.0   \n",
       "151637920.0                           1                         1.0   \n",
       "168509952.0                          12                        12.0   \n",
       "\n",
       "             freq_download_last_14_days  freq_download_last_7_days  \\\n",
       "uid                                                                  \n",
       "167905248.0                         1.0                        NaN   \n",
       "168302176.0                        36.0                        5.0   \n",
       "169027984.0                       165.0                       74.0   \n",
       "168110000.0                         1.0                        1.0   \n",
       "140533488.0                         NaN                        NaN   \n",
       "168434896.0                         1.0                        NaN   \n",
       "151637920.0                         1.0                        1.0   \n",
       "168509952.0                        12.0                       12.0   \n",
       "\n",
       "             freq_download_last_3_days  freq_download_last_1_days  \n",
       "uid                                                                \n",
       "167905248.0                        NaN                        NaN  \n",
       "168302176.0                        4.0                        4.0  \n",
       "169027984.0                       64.0                       60.0  \n",
       "168110000.0                        1.0                        1.0  \n",
       "140533488.0                        NaN                        NaN  \n",
       "168434896.0                        NaN                        NaN  \n",
       "151637920.0                        1.0                        1.0  \n",
       "168509952.0                       12.0                       12.0  "
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "download_feature.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4738, 6)"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "download_feature.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "download_feature.to_csv(\"D:/MusicFile/download_feature_318.csv\", header = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "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>freq_search_last_21_days</th>\n",
       "      <th>freq_search_last_14_days</th>\n",
       "      <th>freq_search_last_7_days</th>\n",
       "      <th>freq_search_last_3_days</th>\n",
       "      <th>freq_search_last_1_days</th>\n",
       "      <th>freq_download_last_21_days</th>\n",
       "      <th>freq_download_last_14_days</th>\n",
       "      <th>freq_download_last_7_days</th>\n",
       "      <th>freq_download_last_3_days</th>\n",
       "      <th>freq_download_last_1_days</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3628</th>\n",
       "      <td>168710656.0</td>\n",
       "      <td>3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>6.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3654</th>\n",
       "      <td>168719200.0</td>\n",
       "      <td>17</td>\n",
       "      <td>14.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130</th>\n",
       "      <td>125439600.0</td>\n",
       "      <td>2</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4513</th>\n",
       "      <td>169007776.0</td>\n",
       "      <td>58</td>\n",
       "      <td>48.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>39.0</td>\n",
       "      <td>35.0</td>\n",
       "      <td>162.0</td>\n",
       "      <td>153.0</td>\n",
       "      <td>152.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>9.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3093</th>\n",
       "      <td>168515184.0</td>\n",
       "      <td>35</td>\n",
       "      <td>10.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2288</th>\n",
       "      <td>168248544.0</td>\n",
       "      <td>8</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>31.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3017</th>\n",
       "      <td>168489712.0</td>\n",
       "      <td>18</td>\n",
       "      <td>13.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2735</th>\n",
       "      <td>168397136.0</td>\n",
       "      <td>15</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>9.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3874</th>\n",
       "      <td>168789456.0</td>\n",
       "      <td>14</td>\n",
       "      <td>12.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1960</th>\n",
       "      <td>168130448.0</td>\n",
       "      <td>90</td>\n",
       "      <td>65.0</td>\n",
       "      <td>37.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>43.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>8.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              uid  freq_search_last_21_days  freq_search_last_14_days  \\\n",
       "3628  168710656.0                         3                       NaN   \n",
       "3654  168719200.0                        17                      14.0   \n",
       "130   125439600.0                         2                       2.0   \n",
       "4513  169007776.0                        58                      48.0   \n",
       "3093  168515184.0                        35                      10.0   \n",
       "2288  168248544.0                         8                       5.0   \n",
       "3017  168489712.0                        18                      13.0   \n",
       "2735  168397136.0                        15                       1.0   \n",
       "3874  168789456.0                        14                      12.0   \n",
       "1960  168130448.0                        90                      65.0   \n",
       "\n",
       "      freq_search_last_7_days  freq_search_last_3_days  \\\n",
       "3628                      NaN                      NaN   \n",
       "3654                     10.0                      8.0   \n",
       "130                       2.0                      NaN   \n",
       "4513                     40.0                     39.0   \n",
       "3093                      8.0                      6.0   \n",
       "2288                      5.0                      5.0   \n",
       "3017                     11.0                      8.0   \n",
       "2735                      NaN                      NaN   \n",
       "3874                      7.0                      1.0   \n",
       "1960                     37.0                     28.0   \n",
       "\n",
       "      freq_search_last_1_days  freq_download_last_21_days  \\\n",
       "3628                      NaN                         6.0   \n",
       "3654                      8.0                        11.0   \n",
       "130                       NaN                         NaN   \n",
       "4513                     35.0                       162.0   \n",
       "3093                      6.0                         4.0   \n",
       "2288                      5.0                        31.0   \n",
       "3017                      8.0                         4.0   \n",
       "2735                      NaN                         9.0   \n",
       "3874                      1.0                         7.0   \n",
       "1960                     28.0                        43.0   \n",
       "\n",
       "      freq_download_last_14_days  freq_download_last_7_days  \\\n",
       "3628                         2.0                        NaN   \n",
       "3654                         8.0                        4.0   \n",
       "130                          NaN                        NaN   \n",
       "4513                       153.0                      152.0   \n",
       "3093                         NaN                        NaN   \n",
       "2288                         2.0                        2.0   \n",
       "3017                         3.0                        3.0   \n",
       "2735                         NaN                        NaN   \n",
       "3874                         5.0                        5.0   \n",
       "1960                         9.0                        8.0   \n",
       "\n",
       "      freq_download_last_3_days  freq_download_last_1_days  \n",
       "3628                        NaN                        NaN  \n",
       "3654                        3.0                        3.0  \n",
       "130                         NaN                        NaN  \n",
       "4513                       10.0                        9.0  \n",
       "3093                        NaN                        NaN  \n",
       "2288                        2.0                        2.0  \n",
       "3017                        3.0                        3.0  \n",
       "2735                        NaN                        NaN  \n",
       "3874                        1.0                        NaN  \n",
       "1960                        8.0                        8.0  "
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#join search and download\n",
    "ds = pd.merge(search_feature, download_feature, left_on = 'uid', right_on = 'uid',  how = 'left')\n",
    "ds.sample(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.5 Recency\n",
    "what is the last show up day of one user in three log.type?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "metadata": {},
   "outputs": [],
   "source": [
    "snapshot_date = pd.to_datetime('2017-04-29')\n",
    "snapshot_date = pd.to_datetime(snapshot_date)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 147,
   "metadata": {},
   "outputs": [],
   "source": [
    "def compute_recency(df, types, snapshot_date):\n",
    "    \n",
    "    last = df.groupby('uid')['date'].agg(['max'])\n",
    "    last['last_' + types] = (snapshot_date - last['max']).apply(lambda x: x.days)\n",
    " \n",
    "    last.drop(columns=['max'], inplace=True)\n",
    "    return last"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Timestamp('2017-04-28 00:00:00')"
      ]
     },
     "execution_count": 159,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#filter search/download\n",
    "\n",
    "sampled_search = sampled_search[sampled_search['date'] <snapshot_date]\n",
    "sampled_download = sampled_download[sampled_download['date'] <snapshot_date]\n",
    "sampled_search['date'].max()\n",
    "sampled_download['date'].max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "metadata": {},
   "outputs": [],
   "source": [
    "recency_play = compute_recency(sample_play, 'play', snapshot_date)\n",
    "recency_search = compute_recency(sampled_search, 'search', snapshot_date)\n",
    "recency_download = compute_recency(sampled_download, 'download', snapshot_date)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "recency <0 means user have some active after 4.29, because we only filter by play not search/download"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "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>last_download</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>uid</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [last_download]\n",
       "Index: []"
      ]
     },
     "execution_count": 162,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "recency_search[recency_search.last_search.isnull()]\n",
    "recency_download[recency_download.last_download.isnull()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 150,
   "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>last_search</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>uid</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>838955.0</th>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2285308.0</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2380836.0</th>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3068886.0</th>\n",
       "      <td>27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3414451.0</th>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           last_search\n",
       "uid                   \n",
       "838955.0            15\n",
       "2285308.0            1\n",
       "2380836.0           30\n",
       "3068886.0           27\n",
       "3414451.0            3"
      ]
     },
     "execution_count": 150,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "recency_search.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "metadata": {},
   "outputs": [],
   "source": [
    "recency_play = recency_play.reset_index(inplace = False)\n",
    "recency_search= recency_search.reset_index(inplace = False)\n",
    "recency_download=recency_download.reset_index(inplace = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.Join play, search, download feature\n",
    "Since our analysis based on play, so if user doesn't have search, download, just plugged in NaN\n",
    "play_feature_all has 8798 user, search has 5211, download has 4738"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.5922937031143441 0.5385314844282791\n"
     ]
    }
   ],
   "source": [
    "print(5211/8798, 4738/8798)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "metadata": {},
   "outputs": [],
   "source": [
    "#reload type_feature\n",
    "\n",
    "play_feature_all = pd.read_csv(\"D:/MusicFile/play_feature_8798_30.csv\")\n",
    "search_feature = pd.read_csv(\"D:/MusicFile/search_feature_318.csv\")\n",
    "download_feature = pd.read_csv(\"D:/MusicFile/download_feature_318.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "play_feature: (8798, 15)\n",
      "search_feature: (5211, 7)\n",
      "download_feature: (4738, 7)\n"
     ]
    }
   ],
   "source": [
    "print(\"play_feature:\",play_feature_all.shape)\n",
    "print(\"search_feature:\",search_feature.shape)\n",
    "print(\"download_feature:\",download_feature.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "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>last_play</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>838955</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>971787</td>\n",
       "      <td>31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1264383</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1901839</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2163514</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       uid  last_play\n",
       "0   838955         13\n",
       "1   971787         31\n",
       "2  1264383         30\n",
       "3  1901839          1\n",
       "4  2163514          2"
      ]
     },
     "execution_count": 131,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "recency_play.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_feature = play_feature_all.merge(search_feature, on = 'uid', how = \"left\") \\\n",
    "                              .merge(download_feature, on = 'uid', how = 'left') \\\n",
    "                              .merge(recency_play, on = 'uid', how = 'left') \\\n",
    "                            .merge(recency_search, on = 'uid', how = 'left') \\\n",
    "            .merge(recency_download, on = 'uid', how = 'left') \\\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "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>freq_play_last_30_days</th>\n",
       "      <th>avg_playtime_last_30_days</th>\n",
       "      <th>freq_play_last_21_days</th>\n",
       "      <th>avg_playtime_last_21_days</th>\n",
       "      <th>freq_play_last_14_days</th>\n",
       "      <th>avg_playtime_last_14_days</th>\n",
       "      <th>freq_play_last_7_days</th>\n",
       "      <th>avg_playtime_last_7_days</th>\n",
       "      <th>freq_play_last_3_days</th>\n",
       "      <th>...</th>\n",
       "      <th>freq_search_last_1_days</th>\n",
       "      <th>freq_download_last_30_days</th>\n",
       "      <th>freq_download_last_21_days</th>\n",
       "      <th>freq_download_last_14_days</th>\n",
       "      <th>freq_download_last_7_days</th>\n",
       "      <th>freq_download_last_3_days</th>\n",
       "      <th>freq_download_last_1_days</th>\n",
       "      <th>last_play</th>\n",
       "      <th>last_search</th>\n",
       "      <th>last_download</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4074</th>\n",
       "      <td>167625392</td>\n",
       "      <td>25.0</td>\n",
       "      <td>34.680000</td>\n",
       "      <td>19.0</td>\n",
       "      <td>15.052632</td>\n",
       "      <td>19.0</td>\n",
       "      <td>15.052632</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>9</td>\n",
       "      <td>9.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3258</th>\n",
       "      <td>155375696</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>57</td>\n",
       "      <td>57.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>250</th>\n",
       "      <td>73131504</td>\n",
       "      <td>399.0</td>\n",
       "      <td>85.766917</td>\n",
       "      <td>397.0</td>\n",
       "      <td>85.675063</td>\n",
       "      <td>397.0</td>\n",
       "      <td>85.675063</td>\n",
       "      <td>397.0</td>\n",
       "      <td>85.675063</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>121.0</td>\n",
       "      <td>121.0</td>\n",
       "      <td>121.0</td>\n",
       "      <td>121.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8095</th>\n",
       "      <td>168839968</td>\n",
       "      <td>659.0</td>\n",
       "      <td>151.295903</td>\n",
       "      <td>284.0</td>\n",
       "      <td>181.566901</td>\n",
       "      <td>170.0</td>\n",
       "      <td>187.847059</td>\n",
       "      <td>109.0</td>\n",
       "      <td>177.559633</td>\n",
       "      <td>61.0</td>\n",
       "      <td>...</td>\n",
       "      <td>17.0</td>\n",
       "      <td>125.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1</td>\n",
       "      <td>13.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6390</th>\n",
       "      <td>168333040</td>\n",
       "      <td>193.0</td>\n",
       "      <td>206.968912</td>\n",
       "      <td>151.0</td>\n",
       "      <td>225.403974</td>\n",
       "      <td>81.0</td>\n",
       "      <td>291.308642</td>\n",
       "      <td>59.0</td>\n",
       "      <td>316.593220</td>\n",
       "      <td>40.0</td>\n",
       "      <td>...</td>\n",
       "      <td>187.0</td>\n",
       "      <td>287.0</td>\n",
       "      <td>286.0</td>\n",
       "      <td>286.0</td>\n",
       "      <td>286.0</td>\n",
       "      <td>253.0</td>\n",
       "      <td>253.0</td>\n",
       "      <td>3</td>\n",
       "      <td>3.0</td>\n",
       "      <td>6.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8187</th>\n",
       "      <td>168872304</td>\n",
       "      <td>553.0</td>\n",
       "      <td>114.088608</td>\n",
       "      <td>267.0</td>\n",
       "      <td>125.816479</td>\n",
       "      <td>152.0</td>\n",
       "      <td>137.335526</td>\n",
       "      <td>71.0</td>\n",
       "      <td>139.154930</td>\n",
       "      <td>11.0</td>\n",
       "      <td>...</td>\n",
       "      <td>10.0</td>\n",
       "      <td>358.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3406</th>\n",
       "      <td>157764832</td>\n",
       "      <td>5.0</td>\n",
       "      <td>123.600000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>29</td>\n",
       "      <td>29.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>185</th>\n",
       "      <td>52033136</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>59</td>\n",
       "      <td>59.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1326</th>\n",
       "      <td>154477856</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>51</td>\n",
       "      <td>56.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>699</th>\n",
       "      <td>152176320</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>51</td>\n",
       "      <td>56.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10 rows × 30 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            uid  freq_play_last_30_days  avg_playtime_last_30_days  \\\n",
       "4074  167625392                    25.0                  34.680000   \n",
       "3258  155375696                     NaN                        NaN   \n",
       "250    73131504                   399.0                  85.766917   \n",
       "8095  168839968                   659.0                 151.295903   \n",
       "6390  168333040                   193.0                 206.968912   \n",
       "8187  168872304                   553.0                 114.088608   \n",
       "3406  157764832                     5.0                 123.600000   \n",
       "185    52033136                     NaN                        NaN   \n",
       "1326  154477856                     NaN                        NaN   \n",
       "699   152176320                     NaN                        NaN   \n",
       "\n",
       "      freq_play_last_21_days  avg_playtime_last_21_days  \\\n",
       "4074                    19.0                  15.052632   \n",
       "3258                     NaN                        NaN   \n",
       "250                    397.0                  85.675063   \n",
       "8095                   284.0                 181.566901   \n",
       "6390                   151.0                 225.403974   \n",
       "8187                   267.0                 125.816479   \n",
       "3406                     NaN                        NaN   \n",
       "185                      NaN                        NaN   \n",
       "1326                     NaN                        NaN   \n",
       "699                      NaN                        NaN   \n",
       "\n",
       "      freq_play_last_14_days  avg_playtime_last_14_days  \\\n",
       "4074                    19.0                  15.052632   \n",
       "3258                     NaN                        NaN   \n",
       "250                    397.0                  85.675063   \n",
       "8095                   170.0                 187.847059   \n",
       "6390                    81.0                 291.308642   \n",
       "8187                   152.0                 137.335526   \n",
       "3406                     NaN                        NaN   \n",
       "185                      NaN                        NaN   \n",
       "1326                     NaN                        NaN   \n",
       "699                      NaN                        NaN   \n",
       "\n",
       "      freq_play_last_7_days  avg_playtime_last_7_days  freq_play_last_3_days  \\\n",
       "4074                    NaN                       NaN                    NaN   \n",
       "3258                    NaN                       NaN                    NaN   \n",
       "250                   397.0                 85.675063                    NaN   \n",
       "8095                  109.0                177.559633                   61.0   \n",
       "6390                   59.0                316.593220                   40.0   \n",
       "8187                   71.0                139.154930                   11.0   \n",
       "3406                    NaN                       NaN                    NaN   \n",
       "185                     NaN                       NaN                    NaN   \n",
       "1326                    NaN                       NaN                    NaN   \n",
       "699                     NaN                       NaN                    NaN   \n",
       "\n",
       "          ...        freq_search_last_1_days  freq_download_last_30_days  \\\n",
       "4074      ...                            2.0                         NaN   \n",
       "3258      ...                            NaN                         NaN   \n",
       "250       ...                            NaN                       121.0   \n",
       "8095      ...                           17.0                       125.0   \n",
       "6390      ...                          187.0                       287.0   \n",
       "8187      ...                           10.0                       358.0   \n",
       "3406      ...                            NaN                         NaN   \n",
       "185       ...                            NaN                         NaN   \n",
       "1326      ...                            NaN                         NaN   \n",
       "699       ...                            NaN                         NaN   \n",
       "\n",
       "      freq_download_last_21_days  freq_download_last_14_days  \\\n",
       "4074                         NaN                         NaN   \n",
       "3258                         NaN                         NaN   \n",
       "250                        121.0                       121.0   \n",
       "8095                        14.0                         8.0   \n",
       "6390                       286.0                       286.0   \n",
       "8187                         8.0                         2.0   \n",
       "3406                         NaN                         NaN   \n",
       "185                          NaN                         NaN   \n",
       "1326                         NaN                         NaN   \n",
       "699                          NaN                         NaN   \n",
       "\n",
       "     freq_download_last_7_days  freq_download_last_3_days  \\\n",
       "4074                       NaN                        NaN   \n",
       "3258                       NaN                        NaN   \n",
       "250                      121.0                        NaN   \n",
       "8095                       8.0                        5.0   \n",
       "6390                     286.0                      253.0   \n",
       "8187                       1.0                        NaN   \n",
       "3406                       NaN                        NaN   \n",
       "185                        NaN                        NaN   \n",
       "1326                       NaN                        NaN   \n",
       "699                        NaN                        NaN   \n",
       "\n",
       "      freq_download_last_1_days  last_play  last_search  last_download  \n",
       "4074                        NaN          9          9.0            NaN  \n",
       "3258                        NaN         57         57.0            NaN  \n",
       "250                         NaN          5          5.0            5.0  \n",
       "8095                        5.0          1         13.0            1.0  \n",
       "6390                      253.0          3          3.0            6.0  \n",
       "8187                        NaN          1          4.0            4.0  \n",
       "3406                        NaN         29         29.0            NaN  \n",
       "185                         NaN         59         59.0            NaN  \n",
       "1326                        NaN         51         56.0            NaN  \n",
       "699                         NaN         51         56.0            NaN  \n",
       "\n",
       "[10 rows x 30 columns]"
      ]
     },
     "execution_count": 189,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_feature.sample(10)    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8798, 30)"
      ]
     },
     "execution_count": 190,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_feature.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 203,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_feature_in = all_feature.drop('Churn',axis = 1).drop('device', axis = 1).drop('last_play', axis = 1)\n",
    "cdp = all_feature[['uid','Churn', 'device', 'last_play']]\n",
    "all_feature_in = all_feature_in.set_index('uid')\n",
    "#all_feature_in.isnull().sum()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "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>Churn</th>\n",
       "      <th>device</th>\n",
       "      <th>last_play</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>838955</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>971787</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "      <td>31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1264383</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1901839</td>\n",
       "      <td>0</td>\n",
       "      <td>ar</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2163514</td>\n",
       "      <td>0</td>\n",
       "      <td>ar</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       uid  Churn device  last_play\n",
       "0   838955      1     ar         13\n",
       "1   971787      1     ar         31\n",
       "2  1264383      1     ar         30\n",
       "3  1901839      0     ar          1\n",
       "4  2163514      0     ar          2"
      ]
     },
     "execution_count": 204,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cdp.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "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>freq_play_last_30_days</th>\n",
       "      <th>avg_playtime_last_30_days</th>\n",
       "      <th>freq_play_last_21_days</th>\n",
       "      <th>avg_playtime_last_21_days</th>\n",
       "      <th>freq_play_last_14_days</th>\n",
       "      <th>avg_playtime_last_14_days</th>\n",
       "      <th>freq_play_last_7_days</th>\n",
       "      <th>avg_playtime_last_7_days</th>\n",
       "      <th>freq_play_last_3_days</th>\n",
       "      <th>avg_playtime_last_3_days</th>\n",
       "      <th>...</th>\n",
       "      <th>freq_search_last_3_days</th>\n",
       "      <th>freq_search_last_1_days</th>\n",
       "      <th>freq_download_last_30_days</th>\n",
       "      <th>freq_download_last_21_days</th>\n",
       "      <th>freq_download_last_14_days</th>\n",
       "      <th>freq_download_last_7_days</th>\n",
       "      <th>freq_download_last_3_days</th>\n",
       "      <th>freq_download_last_1_days</th>\n",
       "      <th>last_search</th>\n",
       "      <th>last_download</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",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></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>838955</th>\n",
       "      <td>12.0</td>\n",
       "      <td>181.750000</td>\n",
       "      <td>5.0</td>\n",
       "      <td>209.400000</td>\n",
       "      <td>2.0</td>\n",
       "      <td>210.500000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>15.0</td>\n",
       "      <td>25.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>971787</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1264383</th>\n",
       "      <td>3.0</td>\n",
       "      <td>87.666667</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1901839</th>\n",
       "      <td>521.0</td>\n",
       "      <td>163.731286</td>\n",
       "      <td>422.0</td>\n",
       "      <td>151.727488</td>\n",
       "      <td>309.0</td>\n",
       "      <td>147.407767</td>\n",
       "      <td>167.0</td>\n",
       "      <td>129.688623</td>\n",
       "      <td>56.0</td>\n",
       "      <td>128.357143</td>\n",
       "      <td>...</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>16.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2163514</th>\n",
       "      <td>74.0</td>\n",
       "      <td>159.500000</td>\n",
       "      <td>74.0</td>\n",
       "      <td>159.500000</td>\n",
       "      <td>74.0</td>\n",
       "      <td>159.500000</td>\n",
       "      <td>20.0</td>\n",
       "      <td>127.300000</td>\n",
       "      <td>19.0</td>\n",
       "      <td>124.105263</td>\n",
       "      <td>...</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>30.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 26 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         freq_play_last_30_days  avg_playtime_last_30_days  \\\n",
       "uid                                                          \n",
       "838955                     12.0                 181.750000   \n",
       "971787                      NaN                        NaN   \n",
       "1264383                     3.0                  87.666667   \n",
       "1901839                   521.0                 163.731286   \n",
       "2163514                    74.0                 159.500000   \n",
       "\n",
       "         freq_play_last_21_days  avg_playtime_last_21_days  \\\n",
       "uid                                                          \n",
       "838955                      5.0                 209.400000   \n",
       "971787                      NaN                        NaN   \n",
       "1264383                     NaN                        NaN   \n",
       "1901839                   422.0                 151.727488   \n",
       "2163514                    74.0                 159.500000   \n",
       "\n",
       "         freq_play_last_14_days  avg_playtime_last_14_days  \\\n",
       "uid                                                          \n",
       "838955                      2.0                 210.500000   \n",
       "971787                      NaN                        NaN   \n",
       "1264383                     NaN                        NaN   \n",
       "1901839                   309.0                 147.407767   \n",
       "2163514                    74.0                 159.500000   \n",
       "\n",
       "         freq_play_last_7_days  avg_playtime_last_7_days  \\\n",
       "uid                                                        \n",
       "838955                     NaN                       NaN   \n",
       "971787                     NaN                       NaN   \n",
       "1264383                    NaN                       NaN   \n",
       "1901839                  167.0                129.688623   \n",
       "2163514                   20.0                127.300000   \n",
       "\n",
       "         freq_play_last_3_days  avg_playtime_last_3_days      ...        \\\n",
       "uid                                                           ...         \n",
       "838955                     NaN                       NaN      ...         \n",
       "971787                     NaN                       NaN      ...         \n",
       "1264383                    NaN                       NaN      ...         \n",
       "1901839                   56.0                128.357143      ...         \n",
       "2163514                   19.0                124.105263      ...         \n",
       "\n",
       "         freq_search_last_3_days  freq_search_last_1_days  \\\n",
       "uid                                                         \n",
       "838955                       NaN                      NaN   \n",
       "971787                       NaN                      NaN   \n",
       "1264383                      NaN                      NaN   \n",
       "1901839                      5.0                      5.0   \n",
       "2163514                      1.0                      1.0   \n",
       "\n",
       "         freq_download_last_30_days  freq_download_last_21_days  \\\n",
       "uid                                                               \n",
       "838955                          1.0                         NaN   \n",
       "971787                          NaN                         NaN   \n",
       "1264383                         NaN                         NaN   \n",
       "1901839                         5.0                         5.0   \n",
       "2163514                         NaN                         NaN   \n",
       "\n",
       "         freq_download_last_14_days  freq_download_last_7_days  \\\n",
       "uid                                                              \n",
       "838955                          NaN                        NaN   \n",
       "971787                          NaN                        NaN   \n",
       "1264383                         NaN                        NaN   \n",
       "1901839                         NaN                        NaN   \n",
       "2163514                         NaN                        NaN   \n",
       "\n",
       "         freq_download_last_3_days  freq_download_last_1_days  last_search  \\\n",
       "uid                                                                          \n",
       "838955                         NaN                        NaN         15.0   \n",
       "971787                         NaN                        NaN          NaN   \n",
       "1264383                        NaN                        NaN          NaN   \n",
       "1901839                        NaN                        NaN          1.0   \n",
       "2163514                        NaN                        NaN         30.0   \n",
       "\n",
       "         last_download  \n",
       "uid                     \n",
       "838955            25.0  \n",
       "971787             NaN  \n",
       "1264383            NaN  \n",
       "1901839           16.0  \n",
       "2163514            NaN  \n",
       "\n",
       "[5 rows x 26 columns]"
      ]
     },
     "execution_count": 205,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_feature_in.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 206,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(432, 26)"
      ]
     },
     "execution_count": 206,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_feature_in[all_feature_in.isnull().all(axis=1)].shape\n",
    "#nan = df[df.isnull().all(axis=1)].index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 207,
   "metadata": {},
   "outputs": [],
   "source": [
    "#if all feature is NaN, drop those user\n",
    "all_feature_in = all_feature_in.dropna(axis = 0, how = 'all')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 208,
   "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>freq_play_last_30_days</th>\n",
       "      <th>avg_playtime_last_30_days</th>\n",
       "      <th>freq_play_last_21_days</th>\n",
       "      <th>avg_playtime_last_21_days</th>\n",
       "      <th>freq_play_last_14_days</th>\n",
       "      <th>avg_playtime_last_14_days</th>\n",
       "      <th>freq_play_last_7_days</th>\n",
       "      <th>avg_playtime_last_7_days</th>\n",
       "      <th>freq_play_last_3_days</th>\n",
       "      <th>avg_playtime_last_3_days</th>\n",
       "      <th>...</th>\n",
       "      <th>freq_search_last_14_days</th>\n",
       "      <th>freq_search_last_7_days</th>\n",
       "      <th>freq_search_last_3_days</th>\n",
       "      <th>freq_search_last_1_days</th>\n",
       "      <th>freq_download_last_30_days</th>\n",
       "      <th>freq_download_last_21_days</th>\n",
       "      <th>freq_download_last_14_days</th>\n",
       "      <th>freq_download_last_7_days</th>\n",
       "      <th>freq_download_last_3_days</th>\n",
       "      <th>freq_download_last_1_days</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",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></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>168777776</th>\n",
       "      <td>22.0</td>\n",
       "      <td>115.545455</td>\n",
       "      <td>16.0</td>\n",
       "      <td>92.562500</td>\n",
       "      <td>9.0</td>\n",
       "      <td>27.555556</td>\n",
       "      <td>9.0</td>\n",
       "      <td>27.555556</td>\n",
       "      <td>9.0</td>\n",
       "      <td>27.555556</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154535568</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168335152</th>\n",
       "      <td>136.0</td>\n",
       "      <td>77.250000</td>\n",
       "      <td>43.0</td>\n",
       "      <td>78.418605</td>\n",
       "      <td>42.0</td>\n",
       "      <td>77.000000</td>\n",
       "      <td>21.0</td>\n",
       "      <td>77.761905</td>\n",
       "      <td>21.0</td>\n",
       "      <td>77.761905</td>\n",
       "      <td>...</td>\n",
       "      <td>64.0</td>\n",
       "      <td>62.0</td>\n",
       "      <td>62.0</td>\n",
       "      <td>62.0</td>\n",
       "      <td>35.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>23.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>20.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168437616</th>\n",
       "      <td>145.0</td>\n",
       "      <td>133.089655</td>\n",
       "      <td>37.0</td>\n",
       "      <td>95.864865</td>\n",
       "      <td>22.0</td>\n",
       "      <td>114.772727</td>\n",
       "      <td>6.0</td>\n",
       "      <td>77.666667</td>\n",
       "      <td>1.0</td>\n",
       "      <td>243.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>18.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154338752</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168948336</th>\n",
       "      <td>883.0</td>\n",
       "      <td>170.419026</td>\n",
       "      <td>489.0</td>\n",
       "      <td>180.586912</td>\n",
       "      <td>364.0</td>\n",
       "      <td>189.793956</td>\n",
       "      <td>150.0</td>\n",
       "      <td>238.420000</td>\n",
       "      <td>62.0</td>\n",
       "      <td>245.338710</td>\n",
       "      <td>...</td>\n",
       "      <td>45.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>59.0</td>\n",
       "      <td>34.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>22.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>20.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>168169760</th>\n",
       "      <td>840.0</td>\n",
       "      <td>227.546429</td>\n",
       "      <td>412.0</td>\n",
       "      <td>218.667476</td>\n",
       "      <td>363.0</td>\n",
       "      <td>209.944904</td>\n",
       "      <td>237.0</td>\n",
       "      <td>232.143460</td>\n",
       "      <td>77.0</td>\n",
       "      <td>215.675325</td>\n",
       "      <td>...</td>\n",
       "      <td>16.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167947616</th>\n",
       "      <td>5.0</td>\n",
       "      <td>183.600000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>8 rows × 24 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           freq_play_last_30_days  avg_playtime_last_30_days  \\\n",
       "uid                                                            \n",
       "168777776                    22.0                 115.545455   \n",
       "154535568                     0.0                   0.000000   \n",
       "168335152                   136.0                  77.250000   \n",
       "168437616                   145.0                 133.089655   \n",
       "154338752                     0.0                   0.000000   \n",
       "168948336                   883.0                 170.419026   \n",
       "168169760                   840.0                 227.546429   \n",
       "167947616                     5.0                 183.600000   \n",
       "\n",
       "           freq_play_last_21_days  avg_playtime_last_21_days  \\\n",
       "uid                                                            \n",
       "168777776                    16.0                  92.562500   \n",
       "154535568                     0.0                   0.000000   \n",
       "168335152                    43.0                  78.418605   \n",
       "168437616                    37.0                  95.864865   \n",
       "154338752                     0.0                   0.000000   \n",
       "168948336                   489.0                 180.586912   \n",
       "168169760                   412.0                 218.667476   \n",
       "167947616                     0.0                   0.000000   \n",
       "\n",
       "           freq_play_last_14_days  avg_playtime_last_14_days  \\\n",
       "uid                                                            \n",
       "168777776                     9.0                  27.555556   \n",
       "154535568                     0.0                   0.000000   \n",
       "168335152                    42.0                  77.000000   \n",
       "168437616                    22.0                 114.772727   \n",
       "154338752                     0.0                   0.000000   \n",
       "168948336                   364.0                 189.793956   \n",
       "168169760                   363.0                 209.944904   \n",
       "167947616                     0.0                   0.000000   \n",
       "\n",
       "           freq_play_last_7_days  avg_playtime_last_7_days  \\\n",
       "uid                                                          \n",
       "168777776                    9.0                 27.555556   \n",
       "154535568                    0.0                  0.000000   \n",
       "168335152                   21.0                 77.761905   \n",
       "168437616                    6.0                 77.666667   \n",
       "154338752                    0.0                  0.000000   \n",
       "168948336                  150.0                238.420000   \n",
       "168169760                  237.0                232.143460   \n",
       "167947616                    0.0                  0.000000   \n",
       "\n",
       "           freq_play_last_3_days  avg_playtime_last_3_days  \\\n",
       "uid                                                          \n",
       "168777776                    9.0                 27.555556   \n",
       "154535568                    0.0                  0.000000   \n",
       "168335152                   21.0                 77.761905   \n",
       "168437616                    1.0                243.000000   \n",
       "154338752                    0.0                  0.000000   \n",
       "168948336                   62.0                245.338710   \n",
       "168169760                   77.0                215.675325   \n",
       "167947616                    0.0                  0.000000   \n",
       "\n",
       "                     ...              freq_search_last_14_days  \\\n",
       "uid                  ...                                         \n",
       "168777776            ...                                   0.0   \n",
       "154535568            ...                                   0.0   \n",
       "168335152            ...                                  64.0   \n",
       "168437616            ...                                  18.0   \n",
       "154338752            ...                                   0.0   \n",
       "168948336            ...                                  45.0   \n",
       "168169760            ...                                  16.0   \n",
       "167947616            ...                                   0.0   \n",
       "\n",
       "           freq_search_last_7_days  freq_search_last_3_days  \\\n",
       "uid                                                           \n",
       "168777776                      0.0                      0.0   \n",
       "154535568                      0.0                      0.0   \n",
       "168335152                     62.0                     62.0   \n",
       "168437616                      5.0                      3.0   \n",
       "154338752                      0.0                      0.0   \n",
       "168948336                     29.0                     25.0   \n",
       "168169760                     13.0                     13.0   \n",
       "167947616                      0.0                      0.0   \n",
       "\n",
       "           freq_search_last_1_days  freq_download_last_30_days  \\\n",
       "uid                                                              \n",
       "168777776                      0.0                         0.0   \n",
       "154535568                      0.0                         0.0   \n",
       "168335152                     62.0                        35.0   \n",
       "168437616                      3.0                        11.0   \n",
       "154338752                      0.0                         0.0   \n",
       "168948336                     25.0                        59.0   \n",
       "168169760                     12.0                         1.0   \n",
       "167947616                      0.0                         2.0   \n",
       "\n",
       "           freq_download_last_21_days  freq_download_last_14_days  \\\n",
       "uid                                                                 \n",
       "168777776                         0.0                         0.0   \n",
       "154535568                         0.0                         0.0   \n",
       "168335152                        24.0                        23.0   \n",
       "168437616                         9.0                         4.0   \n",
       "154338752                         0.0                         0.0   \n",
       "168948336                        34.0                        25.0   \n",
       "168169760                         1.0                         1.0   \n",
       "167947616                         2.0                         2.0   \n",
       "\n",
       "           freq_download_last_7_days  freq_download_last_3_days  \\\n",
       "uid                                                               \n",
       "168777776                        0.0                        0.0   \n",
       "154535568                        0.0                        0.0   \n",
       "168335152                       20.0                       20.0   \n",
       "168437616                        2.0                        0.0   \n",
       "154338752                        0.0                        0.0   \n",
       "168948336                       22.0                       20.0   \n",
       "168169760                        1.0                        1.0   \n",
       "167947616                        2.0                        2.0   \n",
       "\n",
       "           freq_download_last_1_days  \n",
       "uid                                   \n",
       "168777776                        0.0  \n",
       "154535568                        0.0  \n",
       "168335152                       20.0  \n",
       "168437616                        0.0  \n",
       "154338752                        0.0  \n",
       "168948336                       20.0  \n",
       "168169760                        1.0  \n",
       "167947616                        2.0  \n",
       "\n",
       "[8 rows x 24 columns]"
      ]
     },
     "execution_count": 208,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#then fill na with 0/90\n",
    "all_feature_fill0 = all_feature_in.drop('last_search',axis = 1).drop('last_download', axis = 1)\n",
    "all_feature_fill0 = all_feature_fill0.fillna(0)\n",
    "all_feature_fill0.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 211,
   "metadata": {},
   "outputs": [],
   "source": [
    "recency_feature = all_feature_in[['last_search', 'last_download']]\n",
    "recency_feature= recency_feature.reset_index(inplace= False)\n",
    "#all_feature_fill0 = all_feature_fill0.reset_index(inplace = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 210,
   "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>Churn</th>\n",
       "      <th>device</th>\n",
       "      <th>last_play</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>838955</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>971787</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "      <td>31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1264383</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1901839</td>\n",
       "      <td>0</td>\n",
       "      <td>ar</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2163514</td>\n",
       "      <td>0</td>\n",
       "      <td>ar</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       uid  Churn device  last_play\n",
       "0   838955      1     ar         13\n",
       "1   971787      1     ar         31\n",
       "2  1264383      1     ar         30\n",
       "3  1901839      0     ar          1\n",
       "4  2163514      0     ar          2"
      ]
     },
     "execution_count": 210,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#cdp = cdp.reset_index(inplace = False)\n",
    "cdp.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 212,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_feature_new = all_feature_fill0.merge(recency_feature, on = 'uid', how = 'left').fillna(90)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 214,
   "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>freq_play_last_30_days</th>\n",
       "      <th>avg_playtime_last_30_days</th>\n",
       "      <th>freq_play_last_21_days</th>\n",
       "      <th>avg_playtime_last_21_days</th>\n",
       "      <th>freq_play_last_14_days</th>\n",
       "      <th>avg_playtime_last_14_days</th>\n",
       "      <th>freq_play_last_7_days</th>\n",
       "      <th>avg_playtime_last_7_days</th>\n",
       "      <th>freq_play_last_3_days</th>\n",
       "      <th>...</th>\n",
       "      <th>freq_search_last_3_days</th>\n",
       "      <th>freq_search_last_1_days</th>\n",
       "      <th>freq_download_last_30_days</th>\n",
       "      <th>freq_download_last_21_days</th>\n",
       "      <th>freq_download_last_14_days</th>\n",
       "      <th>freq_download_last_7_days</th>\n",
       "      <th>freq_download_last_3_days</th>\n",
       "      <th>freq_download_last_1_days</th>\n",
       "      <th>last_search</th>\n",
       "      <th>last_download</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>838955</td>\n",
       "      <td>12.0</td>\n",
       "      <td>181.750000</td>\n",
       "      <td>5.0</td>\n",
       "      <td>209.400000</td>\n",
       "      <td>2.0</td>\n",
       "      <td>210.500000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>25.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1264383</td>\n",
       "      <td>3.0</td>\n",
       "      <td>87.666667</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>90.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1901839</td>\n",
       "      <td>521.0</td>\n",
       "      <td>163.731286</td>\n",
       "      <td>422.0</td>\n",
       "      <td>151.727488</td>\n",
       "      <td>309.0</td>\n",
       "      <td>147.407767</td>\n",
       "      <td>167.0</td>\n",
       "      <td>129.688623</td>\n",
       "      <td>56.0</td>\n",
       "      <td>...</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>16.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2163514</td>\n",
       "      <td>74.0</td>\n",
       "      <td>159.500000</td>\n",
       "      <td>74.0</td>\n",
       "      <td>159.500000</td>\n",
       "      <td>74.0</td>\n",
       "      <td>159.500000</td>\n",
       "      <td>20.0</td>\n",
       "      <td>127.300000</td>\n",
       "      <td>19.0</td>\n",
       "      <td>...</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>90.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2280694</td>\n",
       "      <td>12.0</td>\n",
       "      <td>140.333333</td>\n",
       "      <td>9.0</td>\n",
       "      <td>130.888889</td>\n",
       "      <td>7.0</td>\n",
       "      <td>95.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>11.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 27 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       uid  freq_play_last_30_days  avg_playtime_last_30_days  \\\n",
       "0   838955                    12.0                 181.750000   \n",
       "1  1264383                     3.0                  87.666667   \n",
       "2  1901839                   521.0                 163.731286   \n",
       "3  2163514                    74.0                 159.500000   \n",
       "4  2280694                    12.0                 140.333333   \n",
       "\n",
       "   freq_play_last_21_days  avg_playtime_last_21_days  freq_play_last_14_days  \\\n",
       "0                     5.0                 209.400000                     2.0   \n",
       "1                     0.0                   0.000000                     0.0   \n",
       "2                   422.0                 151.727488                   309.0   \n",
       "3                    74.0                 159.500000                    74.0   \n",
       "4                     9.0                 130.888889                     7.0   \n",
       "\n",
       "   avg_playtime_last_14_days  freq_play_last_7_days  avg_playtime_last_7_days  \\\n",
       "0                 210.500000                    0.0                  0.000000   \n",
       "1                   0.000000                    0.0                  0.000000   \n",
       "2                 147.407767                  167.0                129.688623   \n",
       "3                 159.500000                   20.0                127.300000   \n",
       "4                  95.000000                    0.0                  0.000000   \n",
       "\n",
       "   freq_play_last_3_days      ...        freq_search_last_3_days  \\\n",
       "0                    0.0      ...                            0.0   \n",
       "1                    0.0      ...                            0.0   \n",
       "2                   56.0      ...                            5.0   \n",
       "3                   19.0      ...                            1.0   \n",
       "4                    0.0      ...                            0.0   \n",
       "\n",
       "   freq_search_last_1_days  freq_download_last_30_days  \\\n",
       "0                      0.0                         1.0   \n",
       "1                      0.0                         0.0   \n",
       "2                      5.0                         5.0   \n",
       "3                      1.0                         0.0   \n",
       "4                      0.0                        14.0   \n",
       "\n",
       "   freq_download_last_21_days  freq_download_last_14_days  \\\n",
       "0                         0.0                         0.0   \n",
       "1                         0.0                         0.0   \n",
       "2                         5.0                         0.0   \n",
       "3                         0.0                         0.0   \n",
       "4                        14.0                         8.0   \n",
       "\n",
       "   freq_download_last_7_days  freq_download_last_3_days  \\\n",
       "0                        0.0                        0.0   \n",
       "1                        0.0                        0.0   \n",
       "2                        0.0                        0.0   \n",
       "3                        0.0                        0.0   \n",
       "4                        5.0                        5.0   \n",
       "\n",
       "   freq_download_last_1_days  last_search  last_download  \n",
       "0                        0.0         15.0           25.0  \n",
       "1                        0.0         90.0           90.0  \n",
       "2                        0.0          1.0           16.0  \n",
       "3                        0.0         30.0           90.0  \n",
       "4                        5.0         90.0           11.0  \n",
       "\n",
       "[5 rows x 27 columns]"
      ]
     },
     "execution_count": 214,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_feature_new.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 215,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8366, 30)"
      ]
     },
     "execution_count": 215,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_feature_all = all_feature_new.merge(cdp, on = 'uid', how = 'inner')\n",
    "all_feature_all.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 216,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['uid', 'freq_play_last_30_days', 'avg_playtime_last_30_days',\n",
       "       'freq_play_last_21_days', 'avg_playtime_last_21_days',\n",
       "       'freq_play_last_14_days', 'avg_playtime_last_14_days',\n",
       "       'freq_play_last_7_days', 'avg_playtime_last_7_days',\n",
       "       'freq_play_last_3_days', 'avg_playtime_last_3_days',\n",
       "       'freq_play_last_1_days', 'avg_playtime_last_1_days',\n",
       "       'freq_search_last_30_days', 'freq_search_last_21_days',\n",
       "       'freq_search_last_14_days', 'freq_search_last_7_days',\n",
       "       'freq_search_last_3_days', 'freq_search_last_1_days',\n",
       "       'freq_download_last_30_days', 'freq_download_last_21_days',\n",
       "       'freq_download_last_14_days', 'freq_download_last_7_days',\n",
       "       'freq_download_last_3_days', 'freq_download_last_1_days', 'last_search',\n",
       "       'last_download', 'Churn', 'device', 'last_play'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 216,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_feature_all.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## One_hot_encoding for device\n",
    "sklearn can not accept numeric variables, for device, either use one-hot encoding or try model without device, save both and try later."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 217,
   "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>freq_play_last_30_days</th>\n",
       "      <th>avg_playtime_last_30_days</th>\n",
       "      <th>freq_play_last_21_days</th>\n",
       "      <th>avg_playtime_last_21_days</th>\n",
       "      <th>freq_play_last_14_days</th>\n",
       "      <th>avg_playtime_last_14_days</th>\n",
       "      <th>freq_play_last_7_days</th>\n",
       "      <th>avg_playtime_last_7_days</th>\n",
       "      <th>freq_play_last_3_days</th>\n",
       "      <th>...</th>\n",
       "      <th>freq_download_last_21_days</th>\n",
       "      <th>freq_download_last_14_days</th>\n",
       "      <th>freq_download_last_7_days</th>\n",
       "      <th>freq_download_last_3_days</th>\n",
       "      <th>freq_download_last_1_days</th>\n",
       "      <th>last_search</th>\n",
       "      <th>last_download</th>\n",
       "      <th>Churn</th>\n",
       "      <th>device</th>\n",
       "      <th>last_play</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3702</th>\n",
       "      <td>167640976</td>\n",
       "      <td>1238.0</td>\n",
       "      <td>148.054927</td>\n",
       "      <td>643.0</td>\n",
       "      <td>151.251944</td>\n",
       "      <td>444.0</td>\n",
       "      <td>176.202703</td>\n",
       "      <td>247.0</td>\n",
       "      <td>197.360324</td>\n",
       "      <td>34.0</td>\n",
       "      <td>...</td>\n",
       "      <td>31.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>both</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1737</th>\n",
       "      <td>154613984</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>56.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "      <td>55</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3612</th>\n",
       "      <td>167618032</td>\n",
       "      <td>12.0</td>\n",
       "      <td>238.500000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "      <td>25</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4505</th>\n",
       "      <td>167881696</td>\n",
       "      <td>73.0</td>\n",
       "      <td>76.479452</td>\n",
       "      <td>40.0</td>\n",
       "      <td>71.025000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>10.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>0</td>\n",
       "      <td>ar</td>\n",
       "      <td>20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>205</th>\n",
       "      <td>71860104</td>\n",
       "      <td>23.0</td>\n",
       "      <td>162.608696</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3376</th>\n",
       "      <td>167457504</td>\n",
       "      <td>1.0</td>\n",
       "      <td>295.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>1</td>\n",
       "      <td>ar</td>\n",
       "      <td>28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4477</th>\n",
       "      <td>167871984</td>\n",
       "      <td>49.0</td>\n",
       "      <td>169.591837</td>\n",
       "      <td>43.0</td>\n",
       "      <td>168.209302</td>\n",
       "      <td>34.0</td>\n",
       "      <td>199.470588</td>\n",
       "      <td>1.0</td>\n",
       "      <td>205.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>1</td>\n",
       "      <td>both</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>769</th>\n",
       "      <td>154426560</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>56.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>1</td>\n",
       "      <td>both</td>\n",
       "      <td>51</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>8 rows × 30 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            uid  freq_play_last_30_days  avg_playtime_last_30_days  \\\n",
       "3702  167640976                  1238.0                 148.054927   \n",
       "1737  154613984                     0.0                   0.000000   \n",
       "3612  167618032                    12.0                 238.500000   \n",
       "4505  167881696                    73.0                  76.479452   \n",
       "205    71860104                    23.0                 162.608696   \n",
       "3376  167457504                     1.0                 295.000000   \n",
       "4477  167871984                    49.0                 169.591837   \n",
       "769   154426560                     0.0                   0.000000   \n",
       "\n",
       "      freq_play_last_21_days  avg_playtime_last_21_days  \\\n",
       "3702                   643.0                 151.251944   \n",
       "1737                     0.0                   0.000000   \n",
       "3612                     0.0                   0.000000   \n",
       "4505                    40.0                  71.025000   \n",
       "205                      0.0                   0.000000   \n",
       "3376                     0.0                   0.000000   \n",
       "4477                    43.0                 168.209302   \n",
       "769                      0.0                   0.000000   \n",
       "\n",
       "      freq_play_last_14_days  avg_playtime_last_14_days  \\\n",
       "3702                   444.0                 176.202703   \n",
       "1737                     0.0                   0.000000   \n",
       "3612                     0.0                   0.000000   \n",
       "4505                     0.0                   0.000000   \n",
       "205                      0.0                   0.000000   \n",
       "3376                     0.0                   0.000000   \n",
       "4477                    34.0                 199.470588   \n",
       "769                      0.0                   0.000000   \n",
       "\n",
       "      freq_play_last_7_days  avg_playtime_last_7_days  freq_play_last_3_days  \\\n",
       "3702                  247.0                197.360324                   34.0   \n",
       "1737                    0.0                  0.000000                    0.0   \n",
       "3612                    0.0                  0.000000                    0.0   \n",
       "4505                    0.0                  0.000000                    0.0   \n",
       "205                     0.0                  0.000000                    0.0   \n",
       "3376                    0.0                  0.000000                    0.0   \n",
       "4477                    1.0                205.000000                    0.0   \n",
       "769                     0.0                  0.000000                    0.0   \n",
       "\n",
       "        ...      freq_download_last_21_days  freq_download_last_14_days  \\\n",
       "3702    ...                            31.0                        20.0   \n",
       "1737    ...                             0.0                         0.0   \n",
       "3612    ...                             0.0                         0.0   \n",
       "4505    ...                            10.0                         0.0   \n",
       "205     ...                             0.0                         0.0   \n",
       "3376    ...                             0.0                         0.0   \n",
       "4477    ...                             4.0                         1.0   \n",
       "769     ...                             0.0                         0.0   \n",
       "\n",
       "      freq_download_last_7_days  freq_download_last_3_days  \\\n",
       "3702                        6.0                        3.0   \n",
       "1737                        0.0                        0.0   \n",
       "3612                        0.0                        0.0   \n",
       "4505                        0.0                        0.0   \n",
       "205                         0.0                        0.0   \n",
       "3376                        0.0                        0.0   \n",
       "4477                        0.0                        0.0   \n",
       "769                         0.0                        0.0   \n",
       "\n",
       "      freq_download_last_1_days  last_search  last_download  Churn  device  \\\n",
       "3702                        2.0          4.0            2.0      0    both   \n",
       "1737                        0.0         56.0           90.0      1      ar   \n",
       "3612                        0.0         90.0           90.0      1      ar   \n",
       "4505                        0.0         20.0           20.0      0      ar   \n",
       "205                         0.0         90.0           90.0      1      ar   \n",
       "3376                        0.0         28.0           90.0      1      ar   \n",
       "4477                        0.0          5.0           11.0      1    both   \n",
       "769                         0.0         56.0           90.0      1    both   \n",
       "\n",
       "      last_play  \n",
       "3702          1  \n",
       "1737         55  \n",
       "3612         25  \n",
       "4505         20  \n",
       "205          30  \n",
       "3376         28  \n",
       "4477          5  \n",
       "769          51  \n",
       "\n",
       "[8 rows x 30 columns]"
      ]
     },
     "execution_count": 217,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_feature_all.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 218,
   "metadata": {},
   "outputs": [],
   "source": [
    "device_dummies = pd.get_dummies(all_feature_all.device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 219,
   "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>ar</th>\n",
       "      <th>both</th>\n",
       "      <th>ip</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>125</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2968</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1464</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1552</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7671</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3604</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2507</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>143</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4823</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1579</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      ar  both  ip\n",
       "125    1     0   0\n",
       "2968   1     0   0\n",
       "1464   1     0   0\n",
       "1552   0     1   0\n",
       "7671   0     1   0\n",
       "3604   1     0   0\n",
       "2507   0     1   0\n",
       "143    1     0   0\n",
       "4823   0     1   0\n",
       "1579   1     0   0"
      ]
     },
     "execution_count": 219,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "device_dummies.sample(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 221,
   "metadata": {},
   "outputs": [],
   "source": [
    "#merge to the all_feature as all_feature_device_dummy\n",
    "all_feature_all['device_ar'], all_feature_all['device_both'], all_feature_all['device_ip'] = device_dummies.ar, device_dummies.both, device_dummies.ip\n",
    "all_feature_dd = all_feature_all.drop('device', 1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 222,
   "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>freq_play_last_30_days</th>\n",
       "      <th>avg_playtime_last_30_days</th>\n",
       "      <th>freq_play_last_21_days</th>\n",
       "      <th>avg_playtime_last_21_days</th>\n",
       "      <th>freq_play_last_14_days</th>\n",
       "      <th>avg_playtime_last_14_days</th>\n",
       "      <th>freq_play_last_7_days</th>\n",
       "      <th>avg_playtime_last_7_days</th>\n",
       "      <th>freq_play_last_3_days</th>\n",
       "      <th>...</th>\n",
       "      <th>freq_download_last_7_days</th>\n",
       "      <th>freq_download_last_3_days</th>\n",
       "      <th>freq_download_last_1_days</th>\n",
       "      <th>last_search</th>\n",
       "      <th>last_download</th>\n",
       "      <th>Churn</th>\n",
       "      <th>last_play</th>\n",
       "      <th>device_ar</th>\n",
       "      <th>device_both</th>\n",
       "      <th>device_ip</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>7244</th>\n",
       "      <td>168719872</td>\n",
       "      <td>140.0</td>\n",
       "      <td>122.014286</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>0</td>\n",
       "      <td>27</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3670</th>\n",
       "      <td>167632128</td>\n",
       "      <td>305.0</td>\n",
       "      <td>228.816393</td>\n",
       "      <td>176.0</td>\n",
       "      <td>225.545455</td>\n",
       "      <td>37.0</td>\n",
       "      <td>260.621622</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>0</td>\n",
       "      <td>13</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6301</th>\n",
       "      <td>168437072</td>\n",
       "      <td>152.0</td>\n",
       "      <td>212.585526</td>\n",
       "      <td>91.0</td>\n",
       "      <td>239.054945</td>\n",
       "      <td>53.0</td>\n",
       "      <td>256.962264</td>\n",
       "      <td>15.0</td>\n",
       "      <td>286.733333</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2941</th>\n",
       "      <td>155769904</td>\n",
       "      <td>6.0</td>\n",
       "      <td>239.000000</td>\n",
       "      <td>5.0</td>\n",
       "      <td>116.600000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>1</td>\n",
       "      <td>20</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6265</th>\n",
       "      <td>168427088</td>\n",
       "      <td>97.0</td>\n",
       "      <td>111.329897</td>\n",
       "      <td>13.0</td>\n",
       "      <td>93.076923</td>\n",
       "      <td>12.0</td>\n",
       "      <td>83.666667</td>\n",
       "      <td>3.0</td>\n",
       "      <td>33.333333</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 32 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            uid  freq_play_last_30_days  avg_playtime_last_30_days  \\\n",
       "7244  168719872                   140.0                 122.014286   \n",
       "3670  167632128                   305.0                 228.816393   \n",
       "6301  168437072                   152.0                 212.585526   \n",
       "2941  155769904                     6.0                 239.000000   \n",
       "6265  168427088                    97.0                 111.329897   \n",
       "\n",
       "      freq_play_last_21_days  avg_playtime_last_21_days  \\\n",
       "7244                     0.0                   0.000000   \n",
       "3670                   176.0                 225.545455   \n",
       "6301                    91.0                 239.054945   \n",
       "2941                     5.0                 116.600000   \n",
       "6265                    13.0                  93.076923   \n",
       "\n",
       "      freq_play_last_14_days  avg_playtime_last_14_days  \\\n",
       "7244                     0.0                   0.000000   \n",
       "3670                    37.0                 260.621622   \n",
       "6301                    53.0                 256.962264   \n",
       "2941                     0.0                   0.000000   \n",
       "6265                    12.0                  83.666667   \n",
       "\n",
       "      freq_play_last_7_days  avg_playtime_last_7_days  freq_play_last_3_days  \\\n",
       "7244                    0.0                  0.000000                    0.0   \n",
       "3670                    0.0                  0.000000                    0.0   \n",
       "6301                   15.0                286.733333                    0.0   \n",
       "2941                    0.0                  0.000000                    0.0   \n",
       "6265                    3.0                 33.333333                    0.0   \n",
       "\n",
       "        ...      freq_download_last_7_days  freq_download_last_3_days  \\\n",
       "7244    ...                            0.0                        0.0   \n",
       "3670    ...                            0.0                        0.0   \n",
       "6301    ...                            0.0                        0.0   \n",
       "2941    ...                            0.0                        0.0   \n",
       "6265    ...                            1.0                        0.0   \n",
       "\n",
       "      freq_download_last_1_days  last_search  last_download  Churn  last_play  \\\n",
       "7244                        0.0         30.0           25.0      0         27   \n",
       "3670                        0.0          6.0           90.0      0         13   \n",
       "6301                        0.0         90.0           28.0      0          4   \n",
       "2941                        0.0         90.0           20.0      1         20   \n",
       "6265                        0.0         14.0            6.0      0          6   \n",
       "\n",
       "      device_ar  device_both  device_ip  \n",
       "7244          1            0          0  \n",
       "3670          1            0          0  \n",
       "6301          1            0          0  \n",
       "2941          1            0          0  \n",
       "6265          0            1          0  \n",
       "\n",
       "[5 rows x 32 columns]"
      ]
     },
     "execution_count": 222,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_feature_dd.sample(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 223,
   "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>freq_play_last_30_days</th>\n",
       "      <th>avg_playtime_last_30_days</th>\n",
       "      <th>freq_play_last_21_days</th>\n",
       "      <th>avg_playtime_last_21_days</th>\n",
       "      <th>freq_play_last_14_days</th>\n",
       "      <th>avg_playtime_last_14_days</th>\n",
       "      <th>freq_play_last_7_days</th>\n",
       "      <th>avg_playtime_last_7_days</th>\n",
       "      <th>freq_play_last_3_days</th>\n",
       "      <th>...</th>\n",
       "      <th>freq_download_last_30_days</th>\n",
       "      <th>freq_download_last_21_days</th>\n",
       "      <th>freq_download_last_14_days</th>\n",
       "      <th>freq_download_last_7_days</th>\n",
       "      <th>freq_download_last_3_days</th>\n",
       "      <th>freq_download_last_1_days</th>\n",
       "      <th>last_search</th>\n",
       "      <th>last_download</th>\n",
       "      <th>Churn</th>\n",
       "      <th>last_play</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>5204</th>\n",
       "      <td>168094304</td>\n",
       "      <td>1444.0</td>\n",
       "      <td>188.889197</td>\n",
       "      <td>854.0</td>\n",
       "      <td>203.939110</td>\n",
       "      <td>447.0</td>\n",
       "      <td>204.825503</td>\n",
       "      <td>102.0</td>\n",
       "      <td>212.294118</td>\n",
       "      <td>53.0</td>\n",
       "      <td>...</td>\n",
       "      <td>36.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3771</th>\n",
       "      <td>167659632</td>\n",
       "      <td>1462.0</td>\n",
       "      <td>214.954172</td>\n",
       "      <td>979.0</td>\n",
       "      <td>232.043922</td>\n",
       "      <td>578.0</td>\n",
       "      <td>224.261246</td>\n",
       "      <td>196.0</td>\n",
       "      <td>219.867347</td>\n",
       "      <td>71.0</td>\n",
       "      <td>...</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5383</th>\n",
       "      <td>168149152</td>\n",
       "      <td>209.0</td>\n",
       "      <td>219.779904</td>\n",
       "      <td>158.0</td>\n",
       "      <td>234.613924</td>\n",
       "      <td>103.0</td>\n",
       "      <td>231.679612</td>\n",
       "      <td>75.0</td>\n",
       "      <td>227.666667</td>\n",
       "      <td>44.0</td>\n",
       "      <td>...</td>\n",
       "      <td>56.0</td>\n",
       "      <td>37.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5502</th>\n",
       "      <td>168184080</td>\n",
       "      <td>677.0</td>\n",
       "      <td>133.642541</td>\n",
       "      <td>506.0</td>\n",
       "      <td>123.756917</td>\n",
       "      <td>295.0</td>\n",
       "      <td>136.847458</td>\n",
       "      <td>119.0</td>\n",
       "      <td>151.806723</td>\n",
       "      <td>64.0</td>\n",
       "      <td>...</td>\n",
       "      <td>167.0</td>\n",
       "      <td>154.0</td>\n",
       "      <td>46.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5423</th>\n",
       "      <td>168160896</td>\n",
       "      <td>367.0</td>\n",
       "      <td>249.220708</td>\n",
       "      <td>262.0</td>\n",
       "      <td>258.954198</td>\n",
       "      <td>107.0</td>\n",
       "      <td>265.336449</td>\n",
       "      <td>63.0</td>\n",
       "      <td>230.523810</td>\n",
       "      <td>14.0</td>\n",
       "      <td>...</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 29 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            uid  freq_play_last_30_days  avg_playtime_last_30_days  \\\n",
       "5204  168094304                  1444.0                 188.889197   \n",
       "3771  167659632                  1462.0                 214.954172   \n",
       "5383  168149152                   209.0                 219.779904   \n",
       "5502  168184080                   677.0                 133.642541   \n",
       "5423  168160896                   367.0                 249.220708   \n",
       "\n",
       "      freq_play_last_21_days  avg_playtime_last_21_days  \\\n",
       "5204                   854.0                 203.939110   \n",
       "3771                   979.0                 232.043922   \n",
       "5383                   158.0                 234.613924   \n",
       "5502                   506.0                 123.756917   \n",
       "5423                   262.0                 258.954198   \n",
       "\n",
       "      freq_play_last_14_days  avg_playtime_last_14_days  \\\n",
       "5204                   447.0                 204.825503   \n",
       "3771                   578.0                 224.261246   \n",
       "5383                   103.0                 231.679612   \n",
       "5502                   295.0                 136.847458   \n",
       "5423                   107.0                 265.336449   \n",
       "\n",
       "      freq_play_last_7_days  avg_playtime_last_7_days  freq_play_last_3_days  \\\n",
       "5204                  102.0                212.294118                   53.0   \n",
       "3771                  196.0                219.867347                   71.0   \n",
       "5383                   75.0                227.666667                   44.0   \n",
       "5502                  119.0                151.806723                   64.0   \n",
       "5423                   63.0                230.523810                   14.0   \n",
       "\n",
       "        ...      freq_download_last_30_days  freq_download_last_21_days  \\\n",
       "5204    ...                            36.0                        18.0   \n",
       "3771    ...                             2.0                         1.0   \n",
       "5383    ...                            56.0                        37.0   \n",
       "5502    ...                           167.0                       154.0   \n",
       "5423    ...                             2.0                         1.0   \n",
       "\n",
       "      freq_download_last_14_days  freq_download_last_7_days  \\\n",
       "5204                        14.0                       10.0   \n",
       "3771                         1.0                        1.0   \n",
       "5383                        10.0                       10.0   \n",
       "5502                        46.0                       21.0   \n",
       "5423                         1.0                        1.0   \n",
       "\n",
       "      freq_download_last_3_days  freq_download_last_1_days  last_search  \\\n",
       "5204                        8.0                        7.0          7.0   \n",
       "3771                        1.0                        1.0          1.0   \n",
       "5383                        7.0                        7.0          6.0   \n",
       "5502                       17.0                       11.0          3.0   \n",
       "5423                        0.0                        0.0          6.0   \n",
       "\n",
       "      last_download  Churn  last_play  \n",
       "5204            2.0      0          1  \n",
       "3771            1.0      0          1  \n",
       "5383            6.0      0          1  \n",
       "5502            2.0      0          1  \n",
       "5423            6.0      0          1  \n",
       "\n",
       "[5 rows x 29 columns]"
      ]
     },
     "execution_count": 223,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#drop device as all_feature_no_d\n",
    "all_feature_nd = all_feature_dd.drop(['device_ar', 'device_ip', 'device_both'],1)\n",
    "all_feature_nd.sample(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# different version of features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "8366\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>freq_play_last_30_days</th>\n",
       "      <th>avg_playtime_last_30_days</th>\n",
       "      <th>freq_play_last_21_days</th>\n",
       "      <th>avg_playtime_last_21_days</th>\n",
       "      <th>freq_play_last_14_days</th>\n",
       "      <th>avg_playtime_last_14_days</th>\n",
       "      <th>freq_play_last_7_days</th>\n",
       "      <th>avg_playtime_last_7_days</th>\n",
       "      <th>freq_play_last_3_days</th>\n",
       "      <th>...</th>\n",
       "      <th>freq_download_last_3_days</th>\n",
       "      <th>freq_download_last_1_days</th>\n",
       "      <th>last_search</th>\n",
       "      <th>last_download</th>\n",
       "      <th>Churn</th>\n",
       "      <th>last_play</th>\n",
       "      <th>device_ar</th>\n",
       "      <th>device_both</th>\n",
       "      <th>device_ip</th>\n",
       "      <th>precent</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>6290</th>\n",
       "      <td>168434864</td>\n",
       "      <td>242.0</td>\n",
       "      <td>136.239669</td>\n",
       "      <td>185.0</td>\n",
       "      <td>111.756757</td>\n",
       "      <td>106.0</td>\n",
       "      <td>117.396226</td>\n",
       "      <td>2.0</td>\n",
       "      <td>172.500000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>1</td>\n",
       "      <td>7</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.589708</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3985</th>\n",
       "      <td>167721888</td>\n",
       "      <td>1243.0</td>\n",
       "      <td>210.911504</td>\n",
       "      <td>737.0</td>\n",
       "      <td>205.166893</td>\n",
       "      <td>567.0</td>\n",
       "      <td>202.523810</td>\n",
       "      <td>465.0</td>\n",
       "      <td>228.395699</td>\n",
       "      <td>176.0</td>\n",
       "      <td>...</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.896268</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5530</th>\n",
       "      <td>168193280</td>\n",
       "      <td>761.0</td>\n",
       "      <td>163.947438</td>\n",
       "      <td>342.0</td>\n",
       "      <td>204.043860</td>\n",
       "      <td>201.0</td>\n",
       "      <td>197.646766</td>\n",
       "      <td>88.0</td>\n",
       "      <td>194.943182</td>\n",
       "      <td>6.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.739399</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6007</th>\n",
       "      <td>168350176</td>\n",
       "      <td>1116.0</td>\n",
       "      <td>161.800179</td>\n",
       "      <td>833.0</td>\n",
       "      <td>166.774310</td>\n",
       "      <td>578.0</td>\n",
       "      <td>177.406574</td>\n",
       "      <td>245.0</td>\n",
       "      <td>191.620408</td>\n",
       "      <td>61.0</td>\n",
       "      <td>...</td>\n",
       "      <td>9.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.686867</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2819</th>\n",
       "      <td>154834224</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>56.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>1</td>\n",
       "      <td>51</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.821584</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5671</th>\n",
       "      <td>168239504</td>\n",
       "      <td>87.0</td>\n",
       "      <td>191.206897</td>\n",
       "      <td>60.0</td>\n",
       "      <td>198.850000</td>\n",
       "      <td>34.0</td>\n",
       "      <td>187.647059</td>\n",
       "      <td>18.0</td>\n",
       "      <td>186.944444</td>\n",
       "      <td>11.0</td>\n",
       "      <td>...</td>\n",
       "      <td>3.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.635749</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8065</th>\n",
       "      <td>168966752</td>\n",
       "      <td>715.0</td>\n",
       "      <td>157.594406</td>\n",
       "      <td>366.0</td>\n",
       "      <td>180.237705</td>\n",
       "      <td>230.0</td>\n",
       "      <td>167.552174</td>\n",
       "      <td>171.0</td>\n",
       "      <td>153.713450</td>\n",
       "      <td>127.0</td>\n",
       "      <td>...</td>\n",
       "      <td>3.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.624820</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5937</th>\n",
       "      <td>168328640</td>\n",
       "      <td>801.0</td>\n",
       "      <td>90.506866</td>\n",
       "      <td>310.0</td>\n",
       "      <td>134.645161</td>\n",
       "      <td>272.0</td>\n",
       "      <td>135.551471</td>\n",
       "      <td>107.0</td>\n",
       "      <td>126.532710</td>\n",
       "      <td>31.0</td>\n",
       "      <td>...</td>\n",
       "      <td>9.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.400042</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>8 rows × 33 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            uid  freq_play_last_30_days  avg_playtime_last_30_days  \\\n",
       "6290  168434864                   242.0                 136.239669   \n",
       "3985  167721888                  1243.0                 210.911504   \n",
       "5530  168193280                   761.0                 163.947438   \n",
       "6007  168350176                  1116.0                 161.800179   \n",
       "2819  154834224                     0.0                   0.000000   \n",
       "5671  168239504                    87.0                 191.206897   \n",
       "8065  168966752                   715.0                 157.594406   \n",
       "5937  168328640                   801.0                  90.506866   \n",
       "\n",
       "      freq_play_last_21_days  avg_playtime_last_21_days  \\\n",
       "6290                   185.0                 111.756757   \n",
       "3985                   737.0                 205.166893   \n",
       "5530                   342.0                 204.043860   \n",
       "6007                   833.0                 166.774310   \n",
       "2819                     0.0                   0.000000   \n",
       "5671                    60.0                 198.850000   \n",
       "8065                   366.0                 180.237705   \n",
       "5937                   310.0                 134.645161   \n",
       "\n",
       "      freq_play_last_14_days  avg_playtime_last_14_days  \\\n",
       "6290                   106.0                 117.396226   \n",
       "3985                   567.0                 202.523810   \n",
       "5530                   201.0                 197.646766   \n",
       "6007                   578.0                 177.406574   \n",
       "2819                     0.0                   0.000000   \n",
       "5671                    34.0                 187.647059   \n",
       "8065                   230.0                 167.552174   \n",
       "5937                   272.0                 135.551471   \n",
       "\n",
       "      freq_play_last_7_days  avg_playtime_last_7_days  freq_play_last_3_days  \\\n",
       "6290                    2.0                172.500000                    0.0   \n",
       "3985                  465.0                228.395699                  176.0   \n",
       "5530                   88.0                194.943182                    6.0   \n",
       "6007                  245.0                191.620408                   61.0   \n",
       "2819                    0.0                  0.000000                    0.0   \n",
       "5671                   18.0                186.944444                   11.0   \n",
       "8065                  171.0                153.713450                  127.0   \n",
       "5937                  107.0                126.532710                   31.0   \n",
       "\n",
       "        ...     freq_download_last_3_days  freq_download_last_1_days  \\\n",
       "6290    ...                           0.0                        0.0   \n",
       "3985    ...                           1.0                        1.0   \n",
       "5530    ...                           0.0                        0.0   \n",
       "6007    ...                           9.0                        9.0   \n",
       "2819    ...                           0.0                        0.0   \n",
       "5671    ...                           3.0                        3.0   \n",
       "8065    ...                           3.0                        3.0   \n",
       "5937    ...                           9.0                        9.0   \n",
       "\n",
       "      last_search  last_download  Churn  last_play  device_ar  device_both  \\\n",
       "6290         11.0           16.0      1          7          0            0   \n",
       "3985          4.0            4.0      0          2          1            0   \n",
       "5530          6.0            8.0      0          1          0            1   \n",
       "6007          1.0            1.0      0          1          0            1   \n",
       "2819         56.0           90.0      1         51          1            0   \n",
       "5671          5.0           14.0      0          1          0            1   \n",
       "8065          1.0            7.0      0          1          0            1   \n",
       "5937          1.0            1.0      0          1          1            0   \n",
       "\n",
       "      device_ip   precent  \n",
       "6290          1  0.589708  \n",
       "3985          0  0.896268  \n",
       "5530          0  0.739399  \n",
       "6007          0  0.686867  \n",
       "2819          0  0.821584  \n",
       "5671          0  0.635749  \n",
       "8065          0  0.624820  \n",
       "5937          0  0.400042  \n",
       "\n",
       "[8 rows x 33 columns]"
      ]
     },
     "execution_count": 228,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#add precentage\n",
    "all_feature_total = all_feature_dd.merge(complete_precent, on = 'uid', how = 'inner')\n",
    "print(len(all_feature_total.uid.unique()))\n",
    "all_feature_total.sample(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 229,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_feature_dd.to_csv(\"D:/MusicFile/all_feature_dd_8366.csv\", header = True, index = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_feature_nd.to_csv(\"D:/MusicFile/all_feature_nd_8366.csv\", header = True, index = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 231,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_feature_total.to_csv(\"D:/MusicFile/all_feature_total_8366.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
}