loan_analysis.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "Collapsed": "true"
   },
   "source": [
    "# data analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "data = pd.read_csv(\"./loan_demand_user_analysis.tsv\",sep='\\t')\n",
    "data.columns = [x.split('.')[1] for x in data.columns]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "Collapsed": "false"
   },
   "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>funds_channel_id</th>\n",
       "      <th>server_date</th>\n",
       "      <th>dsd_market_setup_sw_tag</th>\n",
       "      <th>dsd_loan_page_show_tag</th>\n",
       "      <th>dsd_loan_confirm_button_ck_tag</th>\n",
       "      <th>day_from_first_drip_disburse_succ_time</th>\n",
       "      <th>current_drip_credit_line</th>\n",
       "      <th>current_drip_credit_line_balance</th>\n",
       "      <th>current_drip_credit_line_use_rate</th>\n",
       "      <th>...</th>\n",
       "      <th>curren_drip_overdue_day_max</th>\n",
       "      <th>loan_model_v2_overdue_score</th>\n",
       "      <th>loan_model_v2_fraud_score</th>\n",
       "      <th>last_3_month_hit_pkg_prefix_cnt</th>\n",
       "      <th>last_3_month_hit_pkg_loan_cnt</th>\n",
       "      <th>last_3_month_hit_app_gamble_cnt</th>\n",
       "      <th>last_3_month_fastgulf_passenger_finish_order_actual_cost</th>\n",
       "      <th>last_30_day_fastgulf_passenger_finish_order_actual_cost</th>\n",
       "      <th>last_3_month_category_medicalcare_clinic_orders</th>\n",
       "      <th>last_3_month_category_finance_orders</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>100299</td>\n",
       "      <td>3038.0</td>\n",
       "      <td>2020-07-12</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>172.0</td>\n",
       "      <td>16000.0</td>\n",
       "      <td>6046.76</td>\n",
       "      <td>0.62</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</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",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10071683</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2020-07-11</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</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>2</th>\n",
       "      <td>10078453</td>\n",
       "      <td>3038.0</td>\n",
       "      <td>2020-07-10</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>178.0</td>\n",
       "      <td>12000.0</td>\n",
       "      <td>7032.41</td>\n",
       "      <td>0.41</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</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",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10155339</td>\n",
       "      <td>3027.0</td>\n",
       "      <td>2020-07-15</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>7400.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>1.00</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</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",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10364749</td>\n",
       "      <td>3038.0</td>\n",
       "      <td>2020-07-13</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>81.0</td>\n",
       "      <td>5000.0</td>\n",
       "      <td>490.65</td>\n",
       "      <td>0.90</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</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",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 26 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        uid  funds_channel_id server_date  dsd_market_setup_sw_tag  \\\n",
       "0    100299            3038.0  2020-07-12                        0   \n",
       "1  10071683               NaN  2020-07-11                        0   \n",
       "2  10078453            3038.0  2020-07-10                        0   \n",
       "3  10155339            3027.0  2020-07-15                        1   \n",
       "4  10364749            3038.0  2020-07-13                        0   \n",
       "\n",
       "   dsd_loan_page_show_tag  dsd_loan_confirm_button_ck_tag  \\\n",
       "0                       1                               1   \n",
       "1                       0                               0   \n",
       "2                       1                               1   \n",
       "3                       0                               0   \n",
       "4                       0                               0   \n",
       "\n",
       "   day_from_first_drip_disburse_succ_time  current_drip_credit_line  \\\n",
       "0                                   172.0                   16000.0   \n",
       "1                                     NaN                       NaN   \n",
       "2                                   178.0                   12000.0   \n",
       "3                                    11.0                    7400.0   \n",
       "4                                    81.0                    5000.0   \n",
       "\n",
       "   current_drip_credit_line_balance  current_drip_credit_line_use_rate  ...  \\\n",
       "0                           6046.76                               0.62  ...   \n",
       "1                               NaN                                NaN  ...   \n",
       "2                           7032.41                               0.41  ...   \n",
       "3                              0.00                               1.00  ...   \n",
       "4                            490.65                               0.90  ...   \n",
       "\n",
       "   curren_drip_overdue_day_max  loan_model_v2_overdue_score  \\\n",
       "0                          0.0                          NaN   \n",
       "1                          NaN                          NaN   \n",
       "2                          0.0                          NaN   \n",
       "3                          0.0                          NaN   \n",
       "4                          0.0                          NaN   \n",
       "\n",
       "   loan_model_v2_fraud_score  last_3_month_hit_pkg_prefix_cnt  \\\n",
       "0                        NaN                              NaN   \n",
       "1                        NaN                              NaN   \n",
       "2                        NaN                              NaN   \n",
       "3                        NaN                              NaN   \n",
       "4                        NaN                              NaN   \n",
       "\n",
       "   last_3_month_hit_pkg_loan_cnt  last_3_month_hit_app_gamble_cnt  \\\n",
       "0                            NaN                              NaN   \n",
       "1                            NaN                              NaN   \n",
       "2                            NaN                              NaN   \n",
       "3                            NaN                              NaN   \n",
       "4                            NaN                              NaN   \n",
       "\n",
       "   last_3_month_fastgulf_passenger_finish_order_actual_cost  \\\n",
       "0                                                NaN          \n",
       "1                                                NaN          \n",
       "2                                                NaN          \n",
       "3                                                NaN          \n",
       "4                                                NaN          \n",
       "\n",
       "   last_30_day_fastgulf_passenger_finish_order_actual_cost  \\\n",
       "0                                                NaN         \n",
       "1                                                NaN         \n",
       "2                                                NaN         \n",
       "3                                                NaN         \n",
       "4                                                NaN         \n",
       "\n",
       "   last_3_month_category_medicalcare_clinic_orders  \\\n",
       "0                                              NaN   \n",
       "1                                              NaN   \n",
       "2                                              NaN   \n",
       "3                                              NaN   \n",
       "4                                              NaN   \n",
       "\n",
       "   last_3_month_category_finance_orders  \n",
       "0                                   NaN  \n",
       "1                                   NaN  \n",
       "2                                   NaN  \n",
       "3                                   NaN  \n",
       "4                                   NaN  \n",
       "\n",
       "[5 rows x 26 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['uid', 'funds_channel_id', 'server_date', 'dsd_market_setup_sw_tag',\n",
       "       'dsd_loan_page_show_tag', 'dsd_loan_confirm_button_ck_tag',\n",
       "       'day_from_first_drip_disburse_succ_time', 'current_drip_credit_line',\n",
       "       'current_drip_credit_line_balance', 'current_drip_credit_line_use_rate',\n",
       "       'current_drip_interest_rate', 'last_30_day_drip_loan_num',\n",
       "       'last_3_month_drip_loan_num', 'last_30_day_drip_loan_money',\n",
       "       'last_3_month_drip_loan_money', 'current_drip_in_loan_money',\n",
       "       'curren_drip_overdue_day_max', 'loan_model_v2_overdue_score',\n",
       "       'loan_model_v2_fraud_score', 'last_3_month_hit_pkg_prefix_cnt',\n",
       "       'last_3_month_hit_pkg_loan_cnt', 'last_3_month_hit_app_gamble_cnt',\n",
       "       'last_3_month_fastgulf_passenger_finish_order_actual_cost',\n",
       "       'last_30_day_fastgulf_passenger_finish_order_actual_cost',\n",
       "       'last_3_month_category_medicalcare_clinic_orders',\n",
       "       'last_3_month_category_finance_orders'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "data['current_drip_credit_line_balance_rate'] =1- data['current_drip_credit_line_balance'] / data['current_drip_credit_line']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "data = data[(data['current_drip_credit_line_balance_rate']>=0) & (data['current_drip_credit_line_balance_rate']<=1)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 279,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "data['month_from_first_loan'] = data['day_from_first_drip_disburse_succ_time']//30"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 288,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "def gen_bin_set(data_set, x):\n",
    "#     cuts = np.arange(int(10), 100, int(10))\n",
    "#     cut_old = np.percentile(data_set['current_drip_credit_line_use_rate'], cuts) \n",
    "    bins=10\n",
    "    umax=data_set[x].max()\n",
    "    umin=data_set[x].min()\n",
    "    umax=40\n",
    "    step=(umax-umin)/bins\n",
    "    fre_list=[umin+i*step for i in range(bins+1)]\n",
    "    \n",
    "    br_old = data_set.groupby(pd.cut(data_set[x], fre_list))['dsd_loan_confirm_button_ck_tag'].mean()\n",
    "    df = pd.DataFrame({'confirm_rate': br_old})\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 452,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "data_1 = data[(data['current_drip_credit_line']>20000) & (data['current_drip_credit_line']>150000) ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 453,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3015.0    18381\n",
       "3021.0    10153\n",
       "3027.0     1501\n",
       "3050.0      133\n",
       "3026.0        6\n",
       "3023.0        1\n",
       "Name: funds_channel_id, dtype: int64"
      ]
     },
     "execution_count": 453,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_1['funds_channel_id'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 454,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "df = gen_bin_set(data_1,'month_from_first_loan')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 455,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df['confirm_rate'].plot(kind='bar')\n",
    "plt.title(\"confirm rate\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "data_4 = data[(data['current_drip_credit_line']==10000) & (data['loan_model_v2_overdue_score']>600) ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/user/local/miniconda3/lib/python3.7/site-packages/ipykernel_launcher.py:1: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  \"\"\"Entry point for launching an IPython kernel.\n"
     ]
    }
   ],
   "source": [
    "# data_4['current_drip_interest_rate'] = data_4['current_drip_interest_rate'].fillna(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f2ce8080110>"
      ]
     },
     "execution_count": 182,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "f, ax1 = plt.subplots(1, 1, sharex=True, figsize=(10, 6))\n",
    "\n",
    "sns.distplot(data_4[data_4['dsd_loan_confirm_button_ck_tag'] == 0]['current_drip_interest_rate'],bins=50,color=\"g\",kde_kws={\"label\": \"label=0\"})\n",
    "sns.distplot(data_4[data_4['dsd_loan_confirm_button_ck_tag'] == 1]['current_drip_interest_rate'],bins=50,color=\"r\",kde_kws={\"label\": \"label=1\"})\n",
    "sns.distplot(data_4['current_drip_interest_rate'],bins=100,color=\"y\",kde_kws={\"label\": \"all\"})\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "Collapsed": "true"
   },
   "source": [
    "# Tree"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "import os\n",
    "from sklearn import tree\n",
    "import pydotplus \n",
    "from IPython.display import Image\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from pandas import DataFrame\n",
    "from pandas import Series\n",
    "import random\n",
    "from sklearn.externals import joblib\n",
    "from scipy.stats import ks_2samp\n",
    "from sklearn.metrics import roc_curve,auc\n",
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "def ks(preds,y):\n",
    "    get_ks = lambda preds,y: ks_2samp(preds[y==1], preds[y!=1]).statistic\n",
    "    ks=get_ks(preds,y)\n",
    "    return ks\n",
    "def gini(preds,y):\n",
    "    false_positive_rate, recall, thresholds = roc_curve(y,preds)  \n",
    "    roc_auc = auc(false_positive_rate, recall)\n",
    "    gini=2*roc_auc-1\n",
    "    return gini\n",
    "\n",
    "def binning(col,cut_points,labels=None):\n",
    "    minval=col.min()\n",
    "    maxval=col.max()\n",
    "    breaks_points=[minval]+cut_points+[maxval]\n",
    "    if not labels:\n",
    "        labels=range(len(cut_points)+1)\n",
    "    colBin=pd.cut(col,bins=breaks_points,labels=labels,include_lowest=True)\n",
    "    return colBin"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "#读入数据处理变量\n",
    "gf_sample=pd.read_csv('./loan_demand_user_analysis.tsv',delimiter='\\t')\n",
    "gf_sample.columns=[x.split('.')[1] for x in gf_sample.columns]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "gf_sample['month_from_first_loan'] = gf_sample['day_from_first_drip_disburse_succ_time']//30"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['uid', 'funds_channel_id', 'server_date', 'dsd_market_setup_sw_tag',\n",
       "       'dsd_loan_page_show_tag', 'dsd_loan_confirm_button_ck_tag',\n",
       "       'day_from_first_drip_disburse_succ_time', 'last_taken_time',\n",
       "       'first_use_time', 'last_7_day_drip_taken_coupon_num',\n",
       "       'last_30_day_drip_taken_coupon_num',\n",
       "       'last_3_month_drip_taken_coupon_num',\n",
       "       'last_7_day_drip_taken_type_1_coupon_num',\n",
       "       'last_30_day_drip_taken_type_1_coupon_num',\n",
       "       'last_3_month_drip_taken_type_1_coupon_num',\n",
       "       'last_7_day_drip_taken_type_3_coupon_num',\n",
       "       'last_30_day_drip_taken_type_3_coupon_num',\n",
       "       'last_3_month_drip_taken_type_3_coupon_num',\n",
       "       'last_7_day_drip_taken_type_2_coupon_num',\n",
       "       'last_30_day_drip_taken_type_2_coupon_num',\n",
       "       'last_3_month_drip_taken_type_2_coupon_num',\n",
       "       'last_7_day_drip_use_coupon_num', 'last_30_day_drip_use_coupon_num',\n",
       "       'last_3_month_drip_use_coupon_num', 'avg_use_value_amt',\n",
       "       'last_7_day_drip_use_coupon_amt', 'last_30_day_drip_use_coupon_amt',\n",
       "       'last_3_month_drip_use_coupon_amt', 'his_taken_use_rate',\n",
       "       'current_valid_coupon_num', 'current_drip_credit_line',\n",
       "       'current_drip_credit_line_balance', 'current_drip_credit_line_use_rate',\n",
       "       'current_drip_interest_rate', 'last_30_day_drip_loan_num',\n",
       "       'last_3_month_drip_loan_num', 'last_30_day_drip_loan_money',\n",
       "       'last_3_month_drip_loan_money', 'current_drip_in_loan_money',\n",
       "       'curren_drip_overdue_day_max', 'loan_model_v2_overdue_score',\n",
       "       'loan_model_v2_fraud_score', 'last_3_month_hit_pkg_prefix_cnt',\n",
       "       'last_3_month_hit_pkg_loan_cnt', 'last_3_month_hit_app_gamble_cnt',\n",
       "       'last_3_month_fastgulf_passenger_finish_order_actual_cost',\n",
       "       'last_30_day_fastgulf_passenger_finish_order_actual_cost',\n",
       "       'last_3_month_category_medicalcare_clinic_orders',\n",
       "       'last_3_month_category_finance_orders', 'month_from_first_loan'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gf_sample.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "tmp = gf_sample[[#'dsd_market_setup_sw_tag',\n",
    "       'month_from_first_loan', 'current_drip_credit_line',\n",
    "       'current_drip_credit_line_balance', 'current_drip_credit_line_use_rate','current_drip_in_loan_money',\n",
    "       'current_drip_interest_rate', 'last_30_day_drip_loan_num','last_3_month_drip_loan_num', 'last_30_day_drip_loan_money','last_3_month_drip_loan_money',\n",
    "        'current_valid_coupon_num','his_taken_use_rate',\n",
    "        'last_7_day_drip_taken_coupon_num','last_30_day_drip_taken_coupon_num','last_3_month_drip_taken_coupon_num',\n",
    "        'last_7_day_drip_use_coupon_num','last_30_day_drip_use_coupon_num','last_3_month_drip_use_coupon_num','avg_use_value_amt',\n",
    "        'last_7_day_drip_use_coupon_amt','last_30_day_drip_use_coupon_amt','last_3_month_drip_use_coupon_amt',\n",
    "       'dsd_loan_confirm_button_ck_tag'\n",
    "       ]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "month_from_first_loan                  294370\n",
       "current_drip_credit_line               175844\n",
       "current_drip_credit_line_balance       175844\n",
       "current_drip_credit_line_use_rate      201151\n",
       "current_drip_in_loan_money             325418\n",
       "current_drip_interest_rate             175844\n",
       "last_30_day_drip_loan_num             1693556\n",
       "last_3_month_drip_loan_num            1116695\n",
       "last_30_day_drip_loan_money           1693556\n",
       "last_3_month_drip_loan_money          1116695\n",
       "current_valid_coupon_num               122714\n",
       "his_taken_use_rate                      97342\n",
       "last_7_day_drip_taken_coupon_num        97342\n",
       "last_30_day_drip_taken_coupon_num       97342\n",
       "last_3_month_drip_taken_coupon_num      97342\n",
       "last_7_day_drip_use_coupon_num          97342\n",
       "last_30_day_drip_use_coupon_num         97342\n",
       "last_3_month_drip_use_coupon_num        97342\n",
       "avg_use_value_amt                      129545\n",
       "last_7_day_drip_use_coupon_amt          97659\n",
       "last_30_day_drip_use_coupon_amt         98245\n",
       "last_3_month_drip_use_coupon_amt        98336\n",
       "dsd_loan_confirm_button_ck_tag              0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 82,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [],
   "source": [
    "# tmp['month_from_first_loan'] = tmp['month_from_first_loan'].fillna(10000)\n",
    "# tmp['current_drip_credit_line_use_rate'] = tmp['current_drip_credit_line_use_rate'].fillna(1.0)\n",
    "\n",
    "# cols = [\n",
    "#         'current_drip_credit_line_balance',\n",
    "#         'last_30_day_drip_loan_num','last_3_month_drip_loan_num','last_30_day_drip_loan_money','last_3_month_drip_loan_money','current_drip_in_loan_money',\n",
    "#         'avg_use_value_amt','current_valid_coupon_num','his_taken_use_rate',\n",
    "#        'last_7_day_drip_taken_coupon_num','last_30_day_drip_taken_coupon_num','last_3_month_drip_taken_coupon_num',\n",
    "# #        'last_7_day_drip_taken_type_1_coupon_num','last_30_day_drip_taken_type_1_coupon_num','last_3_month_drip_taken_type_1_coupon_num',\n",
    "# #        'last_7_day_drip_taken_type_2_coupon_num','last_30_day_drip_taken_type_1_coupon_num','last_3_month_drip_taken_type_1_coupon_num',\n",
    "# #        'last_7_day_drip_taken_type_3_coupon_num','last_30_day_drip_taken_type_3_coupon_num','last_3_month_drip_taken_type_3_coupon_num',\n",
    "#        'last_7_day_drip_use_coupon_num','last_30_day_drip_use_coupon_num','last_3_month_drip_use_coupon_num',\n",
    "#        'last_7_day_drip_use_coupon_amt','last_30_day_drip_use_coupon_amt','last_3_month_drip_use_coupon_amt'\n",
    "#        ]\n",
    "# for col in cols:\n",
    "#     tmp[col] = tmp[col].fillna(0)\n",
    "tmp = tmp.dropna()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 822096 entries, 2 to 2565789\n",
      "Data columns (total 23 columns):\n",
      "month_from_first_loan                 822096 non-null float64\n",
      "current_drip_credit_line              822096 non-null float64\n",
      "current_drip_credit_line_balance      822096 non-null float64\n",
      "current_drip_credit_line_use_rate     822096 non-null float64\n",
      "current_drip_in_loan_money            822096 non-null float64\n",
      "current_drip_interest_rate            822096 non-null float64\n",
      "last_30_day_drip_loan_num             822096 non-null float64\n",
      "last_3_month_drip_loan_num            822096 non-null float64\n",
      "last_30_day_drip_loan_money           822096 non-null float64\n",
      "last_3_month_drip_loan_money          822096 non-null float64\n",
      "current_valid_coupon_num              822096 non-null float64\n",
      "his_taken_use_rate                    822096 non-null float64\n",
      "last_7_day_drip_taken_coupon_num      822096 non-null float64\n",
      "last_30_day_drip_taken_coupon_num     822096 non-null float64\n",
      "last_3_month_drip_taken_coupon_num    822096 non-null float64\n",
      "last_7_day_drip_use_coupon_num        822096 non-null float64\n",
      "last_30_day_drip_use_coupon_num       822096 non-null float64\n",
      "last_3_month_drip_use_coupon_num      822096 non-null float64\n",
      "avg_use_value_amt                     822096 non-null float64\n",
      "last_7_day_drip_use_coupon_amt        822096 non-null float64\n",
      "last_30_day_drip_use_coupon_amt       822096 non-null float64\n",
      "last_3_month_drip_use_coupon_amt      822096 non-null float64\n",
      "dsd_loan_confirm_button_ck_tag        822096 non-null int64\n",
      "dtypes: float64(22), int64(1)\n",
      "memory usage: 150.5 MB\n"
     ]
    }
   ],
   "source": [
    "tmp.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "month_from_first_loan                 0\n",
       "current_drip_credit_line              0\n",
       "current_drip_credit_line_balance      0\n",
       "current_drip_credit_line_use_rate     0\n",
       "current_drip_in_loan_money            0\n",
       "current_drip_interest_rate            0\n",
       "last_30_day_drip_loan_num             0\n",
       "last_3_month_drip_loan_num            0\n",
       "last_30_day_drip_loan_money           0\n",
       "last_3_month_drip_loan_money          0\n",
       "current_valid_coupon_num              0\n",
       "his_taken_use_rate                    0\n",
       "last_7_day_drip_taken_coupon_num      0\n",
       "last_30_day_drip_taken_coupon_num     0\n",
       "last_3_month_drip_taken_coupon_num    0\n",
       "last_7_day_drip_use_coupon_num        0\n",
       "last_30_day_drip_use_coupon_num       0\n",
       "last_3_month_drip_use_coupon_num      0\n",
       "avg_use_value_amt                     0\n",
       "last_7_day_drip_use_coupon_amt        0\n",
       "last_30_day_drip_use_coupon_amt       0\n",
       "last_3_month_drip_use_coupon_amt      0\n",
       "dsd_loan_confirm_button_ck_tag        0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "x=tmp[['month_from_first_loan', 'current_drip_credit_line','current_drip_credit_line_balance', 'current_drip_credit_line_use_rate',\n",
    "       'current_drip_in_loan_money',\n",
    "       'current_drip_interest_rate', 'last_30_day_drip_loan_num','last_3_month_drip_loan_num', 'last_30_day_drip_loan_money','last_3_month_drip_loan_money',\n",
    "        'current_valid_coupon_num','his_taken_use_rate',\n",
    "        'last_7_day_drip_taken_coupon_num',#'last_30_day_drip_taken_coupon_num','last_3_month_drip_taken_coupon_num',\n",
    "        'last_7_day_drip_use_coupon_num',#'last_30_day_drip_use_coupon_num','last_3_month_drip_use_coupon_num','avg_use_value_amt',\n",
    "        'last_7_day_drip_use_coupon_amt',#'last_30_day_drip_use_coupon_amt','last_3_month_drip_use_coupon_amt'\n",
    "       ]]\n",
    "y=tmp[['dsd_loan_confirm_button_ck_tag']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.3,random_state=0)\n",
    "# x_train = x_train.fillna(0)\n",
    "# x_test = x_test.fillna(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf = tree.DecisionTreeClassifier(criterion='gini', splitter='best', max_depth=4, min_samples_split=2, min_samples_leaf=1000, min_weight_fraction_leaf=0.0, max_features=None, random_state=None, max_leaf_nodes=None, class_weight=None)\n",
    "clf = clf.fit(x_train,y_train)\n",
    "\n",
    "dot_data = tree.export_graphviz(clf, out_file=None,\n",
    "                     feature_names=x_train.columns,   \n",
    "                     filled=True, rounded=True,  \n",
    "                     special_characters=True,proportion=True) \n",
    "graph = pydotplus.graph_from_dot_data(dot_data) \n",
    "\n",
    "graph.write_pdf(\"loan_demand_tree1.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/user/local/miniconda3/lib/python3.7/site-packages/ipykernel_launcher.py:1: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  \"\"\"Entry point for launching an IPython kernel.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[7, 4, 22, 12, 11, 5, 27, 30, 23, 20, 8, 26, 19, 14, 29, 15]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train['nodes'] = clf.apply(x_train)\n",
    "nodes = x_train.nodes.unique().tolist()\n",
    "print(nodes)\n",
    "\n",
    "x_train = pd.concat([x_train,y_train],axis=1)\n",
    "x_train = x_train[(x_train['nodes']!=4)&(x_train['nodes']!=5)&\n",
    "                 (x_train['nodes']!=7)&(x_train['nodes']!=8)&\n",
    "                 (x_train['nodes']!=19)&(x_train['nodes']!=20)].drop(['nodes'],axis=1)\n",
    "y_train = x_train['dsd_loan_confirm_button_ck_tag']\n",
    "x_train = x_train.drop('dsd_loan_confirm_button_ck_tag',axis=1)\n",
    "\n",
    "clf = tree.DecisionTreeClassifier(criterion='gini', splitter='best', max_depth=4, min_samples_split=2, min_samples_leaf=1000, min_weight_fraction_leaf=0.0, max_features=None, random_state=None, max_leaf_nodes=None, class_weight=None)\n",
    "clf = clf.fit(x_train,y_train)\n",
    "\n",
    "dot_data = tree.export_graphviz(clf, out_file=None,\n",
    "                     feature_names=x_train.columns,   \n",
    "                     filled=True, rounded=True,  \n",
    "                     special_characters=True,proportion=True) \n",
    "graph = pydotplus.graph_from_dot_data(dot_data) \n",
    "\n",
    "graph.write_pdf(\"loan_demand_tree2.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[19, 8, 5, 23, 27, 22, 20, 26, 29, 30, 4, 11, 14, 7, 15, 12]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 95,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train['nodes'] = clf.apply(x_train)\n",
    "nodes = x_train.nodes.unique().tolist()\n",
    "print(nodes)\n",
    "\n",
    "x_train = pd.concat([x_train,y_train],axis=1)\n",
    "x_train = x_train[(x_train['nodes']!=4)&(x_train['nodes']!=5)&\n",
    "                 (x_train['nodes']!=7)&(x_train['nodes']!=8)&\n",
    "                 (x_train['nodes']!=26)].drop(['nodes'],axis=1)\n",
    "y_train = x_train['dsd_loan_confirm_button_ck_tag']\n",
    "x_train = x_train.drop('dsd_loan_confirm_button_ck_tag',axis=1)\n",
    "\n",
    "clf = tree.DecisionTreeClassifier(criterion='gini', splitter='best', max_depth=4, min_samples_split=2, min_samples_leaf=1000, min_weight_fraction_leaf=0.0, max_features=None, random_state=None, max_leaf_nodes=None, class_weight=None)\n",
    "clf = clf.fit(x_train,y_train)\n",
    "\n",
    "dot_data = tree.export_graphviz(clf, out_file=None,\n",
    "                     feature_names=x_train.columns,   \n",
    "                     filled=True, rounded=True,  \n",
    "                     special_characters=True,proportion=True) \n",
    "graph = pydotplus.graph_from_dot_data(dot_data) \n",
    "\n",
    "graph.write_pdf(\"loan_demand_tree3.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": [
    "from sklearn import metrics\n",
    "def print_result(y_score, y_test, thr=0.5):\n",
    "    fpr, tpr, thresholds = metrics.roc_curve(y_test, y_score, pos_label = 1)\n",
    "    print('auc:{}'.format(metrics.auc(fpr, tpr)))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc:0.5314517835746999\n",
      "auc:0.5309684363956242\n"
     ]
    }
   ],
   "source": [
    "y_score_train = clf.predict(x_train)\n",
    "print_result(y_score_train, y_train)\n",
    "y_score_test = clf.predict(x_test)\n",
    "print_result(y_score_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "Collapsed": "false"
   },
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}