AMF Data Challenge.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "6e620d99",
   "metadata": {},
   "source": [
    "# <center><u><b>PROJET BIG DATA - WHO ARE THE HFT?</b></u></center>\n",
    "### <center><b>AMF CHALLENGE</b></center>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3f8e2e14",
   "metadata": {},
   "source": [
    "### <u>1- Importation des librairies<u>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 570,
   "id": "ef481435",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd \n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns \n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split \n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.feature_selection import SelectFromModel\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.discriminant_analysis import LinearDiscriminantAnalysis\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.naive_bayes import GaussianNB"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "524b58f5",
   "metadata": {},
   "source": [
    "### <u>2- Importation de la base de données et regroupement de Y et X <u>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 571,
   "id": "e9617dcd",
   "metadata": {
    "scrolled": 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>Trader</th>\n",
       "      <th>type</th>\n",
       "      <th>OTR</th>\n",
       "      <th>OCR</th>\n",
       "      <th>OMR</th>\n",
       "      <th>min_time_two_events</th>\n",
       "      <th>mean_time_two_events</th>\n",
       "      <th>10_p_time_two_events</th>\n",
       "      <th>med_time_two_events</th>\n",
       "      <th>25_p_time_two_events</th>\n",
       "      <th>...</th>\n",
       "      <th>min_dt_TV1_TV2</th>\n",
       "      <th>mean_dt_TV1_TV2</th>\n",
       "      <th>med_dt_TV1_TV2</th>\n",
       "      <th>min_dt_TV1_TV3</th>\n",
       "      <th>mean_dt_TV1_TV3</th>\n",
       "      <th>med_dt_TV1_TV3</th>\n",
       "      <th>min_dt_TV1_TV4</th>\n",
       "      <th>mean_dt_TV1_TV4</th>\n",
       "      <th>med_dt_TV1_TV4</th>\n",
       "      <th>NbSecondWithAtLeatOneTrade</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Trader_285</td>\n",
       "      <td>MIX</td>\n",
       "      <td>1.428571</td>\n",
       "      <td>10.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.000</td>\n",
       "      <td>2.013731</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.018</td>\n",
       "      <td>420.371000</td>\n",
       "      <td>51.083</td>\n",
       "      <td>0.017</td>\n",
       "      <td>17.039333</td>\n",
       "      <td>0.018</td>\n",
       "      <td>0.018</td>\n",
       "      <td>17.039667</td>\n",
       "      <td>0.018</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Trader_285</td>\n",
       "      <td>MIX</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>21258.127</td>\n",
       "      <td>21258.126953</td>\n",
       "      <td>21258.127</td>\n",
       "      <td>21258.127000</td>\n",
       "      <td>21258.127</td>\n",
       "      <td>...</td>\n",
       "      <td>0.021</td>\n",
       "      <td>13.355429</td>\n",
       "      <td>0.644</td>\n",
       "      <td>0.021</td>\n",
       "      <td>20.416500</td>\n",
       "      <td>0.021</td>\n",
       "      <td>0.021</td>\n",
       "      <td>20.416500</td>\n",
       "      <td>0.021</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Trader_285</td>\n",
       "      <td>MIX</td>\n",
       "      <td>1.500000</td>\n",
       "      <td>7.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.000</td>\n",
       "      <td>4269.648575</td>\n",
       "      <td>0.000</td>\n",
       "      <td>651.518100</td>\n",
       "      <td>0.000</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>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Trader_285</td>\n",
       "      <td>MIX</td>\n",
       "      <td>1.050633</td>\n",
       "      <td>83.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.000</td>\n",
       "      <td>1.728176</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000435</td>\n",
       "      <td>0.000</td>\n",
       "      <td>...</td>\n",
       "      <td>65.660</td>\n",
       "      <td>65.660000</td>\n",
       "      <td>65.660</td>\n",
       "      <td>65.660</td>\n",
       "      <td>65.660000</td>\n",
       "      <td>65.660</td>\n",
       "      <td>65.660</td>\n",
       "      <td>65.660000</td>\n",
       "      <td>65.660</td>\n",
       "      <td>82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Trader_285</td>\n",
       "      <td>MIX</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>21261.256</td>\n",
       "      <td>21261.255859</td>\n",
       "      <td>21261.256</td>\n",
       "      <td>21261.256000</td>\n",
       "      <td>21261.256</td>\n",
       "      <td>...</td>\n",
       "      <td>0.020</td>\n",
       "      <td>2.687125</td>\n",
       "      <td>0.021</td>\n",
       "      <td>0.020</td>\n",
       "      <td>3.676200</td>\n",
       "      <td>0.021</td>\n",
       "      <td>0.020</td>\n",
       "      <td>16.099333</td>\n",
       "      <td>0.021</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 37 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       Trader type       OTR   OCR  OMR  min_time_two_events  \\\n",
       "0  Trader_285  MIX  1.428571  10.0  NaN                0.000   \n",
       "1  Trader_285  MIX       NaN   2.0  NaN            21258.127   \n",
       "2  Trader_285  MIX  1.500000   7.0  NaN                0.000   \n",
       "3  Trader_285  MIX  1.050633  83.0  NaN                0.000   \n",
       "4  Trader_285  MIX       NaN   2.0  NaN            21261.256   \n",
       "\n",
       "   mean_time_two_events  10_p_time_two_events  med_time_two_events  \\\n",
       "0              2.013731                 0.000             0.000000   \n",
       "1          21258.126953             21258.127         21258.127000   \n",
       "2           4269.648575                 0.000           651.518100   \n",
       "3              1.728176                 0.000             0.000435   \n",
       "4          21261.255859             21261.256         21261.256000   \n",
       "\n",
       "   25_p_time_two_events  ...  min_dt_TV1_TV2  mean_dt_TV1_TV2  med_dt_TV1_TV2  \\\n",
       "0                 0.000  ...           0.018       420.371000          51.083   \n",
       "1             21258.127  ...           0.021        13.355429           0.644   \n",
       "2                 0.000  ...             NaN              NaN             NaN   \n",
       "3                 0.000  ...          65.660        65.660000          65.660   \n",
       "4             21261.256  ...           0.020         2.687125           0.021   \n",
       "\n",
       "   min_dt_TV1_TV3  mean_dt_TV1_TV3  med_dt_TV1_TV3  min_dt_TV1_TV4  \\\n",
       "0           0.017        17.039333           0.018           0.018   \n",
       "1           0.021        20.416500           0.021           0.021   \n",
       "2             NaN              NaN             NaN             NaN   \n",
       "3          65.660        65.660000          65.660          65.660   \n",
       "4           0.020         3.676200           0.021           0.020   \n",
       "\n",
       "   mean_dt_TV1_TV4  med_dt_TV1_TV4  NbSecondWithAtLeatOneTrade  \n",
       "0        17.039667           0.018                          52  \n",
       "1        20.416500           0.021                          14  \n",
       "2              NaN             NaN                           1  \n",
       "3        65.660000          65.660                          82  \n",
       "4        16.099333           0.021                          12  \n",
       "\n",
       "[5 rows x 37 columns]"
      ]
     },
     "execution_count": 571,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_x=pd.read_csv(\"AMF_train_X.csv\")\n",
    "df_y=pd.read_csv(\"AMF_train_Y.csv\")\n",
    "df=pd.merge(df_y,df_x,on=\"Trader\")\n",
    "df1=df.drop(['Index','Share','Day'],axis=1) # ces variables ne servent à rien ( le num du trader est aussi inutil mais on va\n",
    "                                            # le garder parce qu'on en aura besoin pour le regroupement )\n",
    "df1.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5883a30b",
   "metadata": {},
   "source": [
    "### <u>2- Train/Test Split <u> "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d0f6cebe",
   "metadata": {},
   "source": [
    "Nous allons diviser le dataset avant de faire le nettoyage des données pour éviter le data leakage : le nettoyage sera fait \n",
    "d'abord sur le training set et les valeurs utilisées pour le remplacement des valeurs manquantes seront après utilisées dans le \n",
    "nettoyage du testing set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 572,
   "id": "c855d9f8",
   "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>Trader</th>\n",
       "      <th>type</th>\n",
       "      <th>OTR</th>\n",
       "      <th>OCR</th>\n",
       "      <th>OMR</th>\n",
       "      <th>min_time_two_events</th>\n",
       "      <th>mean_time_two_events</th>\n",
       "      <th>10_p_time_two_events</th>\n",
       "      <th>med_time_two_events</th>\n",
       "      <th>25_p_time_two_events</th>\n",
       "      <th>...</th>\n",
       "      <th>min_dt_TV1_TV2</th>\n",
       "      <th>mean_dt_TV1_TV2</th>\n",
       "      <th>med_dt_TV1_TV2</th>\n",
       "      <th>min_dt_TV1_TV3</th>\n",
       "      <th>mean_dt_TV1_TV3</th>\n",
       "      <th>med_dt_TV1_TV3</th>\n",
       "      <th>min_dt_TV1_TV4</th>\n",
       "      <th>mean_dt_TV1_TV4</th>\n",
       "      <th>med_dt_TV1_TV4</th>\n",
       "      <th>NbSecondWithAtLeatOneTrade</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>62733</th>\n",
       "      <td>Trader_380</td>\n",
       "      <td>NON HFT</td>\n",
       "      <td>4.125000</td>\n",
       "      <td>24.000000</td>\n",
       "      <td>20.307692</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>6003.961542</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>176.227870</td>\n",
       "      <td>6.569422</td>\n",
       "      <td>...</td>\n",
       "      <td>1053.639566</td>\n",
       "      <td>1053.639566</td>\n",
       "      <td>1053.639566</td>\n",
       "      <td>1053.639566</td>\n",
       "      <td>1053.639566</td>\n",
       "      <td>1053.639566</td>\n",
       "      <td>1053.639566</td>\n",
       "      <td>1053.639566</td>\n",
       "      <td>1053.639566</td>\n",
       "      <td>45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50820</th>\n",
       "      <td>Trader_105</td>\n",
       "      <td>NON HFT</td>\n",
       "      <td>3.809524</td>\n",
       "      <td>40.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1704.873301</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</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>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>93570</th>\n",
       "      <td>Trader_35</td>\n",
       "      <td>HFT</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>47.766374</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>0.000008</td>\n",
       "      <td>0.000007</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>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48863</th>\n",
       "      <td>Trader_329</td>\n",
       "      <td>MIX</td>\n",
       "      <td>38.200000</td>\n",
       "      <td>2.139795</td>\n",
       "      <td>1464.333333</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>68.687286</td>\n",
       "      <td>1.204900</td>\n",
       "      <td>19.578550</td>\n",
       "      <td>5.569770</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000004</td>\n",
       "      <td>45.908919</td>\n",
       "      <td>0.463846</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>87.689266</td>\n",
       "      <td>19.777873</td>\n",
       "      <td>0.000004</td>\n",
       "      <td>51.189014</td>\n",
       "      <td>7.931105</td>\n",
       "      <td>246</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>98266</th>\n",
       "      <td>Trader_451</td>\n",
       "      <td>MIX</td>\n",
       "      <td>131.000000</td>\n",
       "      <td>3.184765</td>\n",
       "      <td>2.795164</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>45.247940</td>\n",
       "      <td>0.023045</td>\n",
       "      <td>12.952679</td>\n",
       "      <td>1.568145</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>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 37 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           Trader     type         OTR        OCR          OMR  \\\n",
       "62733  Trader_380  NON HFT    4.125000  24.000000    20.307692   \n",
       "50820  Trader_105  NON HFT    3.809524  40.000000          NaN   \n",
       "93570   Trader_35      HFT         NaN   3.000000          NaN   \n",
       "48863  Trader_329      MIX   38.200000   2.139795  1464.333333   \n",
       "98266  Trader_451      MIX  131.000000   3.184765     2.795164   \n",
       "\n",
       "       min_time_two_events  mean_time_two_events  10_p_time_two_events  \\\n",
       "62733             0.000000           6003.961542              0.000000   \n",
       "50820             0.000000           1704.873301              0.000000   \n",
       "93570             0.000006             47.766374              0.000006   \n",
       "48863             0.000000             68.687286              1.204900   \n",
       "98266             0.000000             45.247940              0.023045   \n",
       "\n",
       "       med_time_two_events  25_p_time_two_events  ...  min_dt_TV1_TV2  \\\n",
       "62733           176.227870              6.569422  ...     1053.639566   \n",
       "50820             0.000000              0.000000  ...             NaN   \n",
       "93570             0.000008              0.000007  ...             NaN   \n",
       "48863            19.578550              5.569770  ...        0.000004   \n",
       "98266            12.952679              1.568145  ...             NaN   \n",
       "\n",
       "       mean_dt_TV1_TV2  med_dt_TV1_TV2  min_dt_TV1_TV3  mean_dt_TV1_TV3  \\\n",
       "62733      1053.639566     1053.639566     1053.639566      1053.639566   \n",
       "50820              NaN             NaN             NaN              NaN   \n",
       "93570              NaN             NaN             NaN              NaN   \n",
       "48863        45.908919        0.463846        0.000002        87.689266   \n",
       "98266              NaN             NaN             NaN              NaN   \n",
       "\n",
       "       med_dt_TV1_TV3  min_dt_TV1_TV4  mean_dt_TV1_TV4  med_dt_TV1_TV4  \\\n",
       "62733     1053.639566     1053.639566      1053.639566     1053.639566   \n",
       "50820             NaN             NaN              NaN             NaN   \n",
       "93570             NaN             NaN              NaN             NaN   \n",
       "48863       19.777873        0.000004        51.189014        7.931105   \n",
       "98266             NaN             NaN              NaN             NaN   \n",
       "\n",
       "       NbSecondWithAtLeatOneTrade  \n",
       "62733                          45  \n",
       "50820                          12  \n",
       "93570                          12  \n",
       "48863                         246  \n",
       "98266                           7  \n",
       "\n",
       "[5 rows x 37 columns]"
      ]
     },
     "execution_count": 572,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#split train/test\n",
    "df_train,df_test=train_test_split(df1,test_size=0.25, random_state=1,shuffle=True ) \n",
    "df_train.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "052be5a1",
   "metadata": {},
   "source": [
    "### <u>3- Data Cleaning <u>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 573,
   "id": "d226b838",
   "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>nbre de NaN</th>\n",
       "      <th>% des NaN</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>OMR</th>\n",
       "      <td>35710</td>\n",
       "      <td>34.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>med_dt_TV1_TV3</th>\n",
       "      <td>18974</td>\n",
       "      <td>18.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean_dt_TV1_TV3</th>\n",
       "      <td>18974</td>\n",
       "      <td>18.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min_dt_TV1_TV3</th>\n",
       "      <td>18974</td>\n",
       "      <td>18.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>med_dt_TV1_TV2</th>\n",
       "      <td>18917</td>\n",
       "      <td>18.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean_dt_TV1_TV2</th>\n",
       "      <td>18917</td>\n",
       "      <td>18.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min_dt_TV1_TV2</th>\n",
       "      <td>18917</td>\n",
       "      <td>18.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>med_dt_TV1_TV4</th>\n",
       "      <td>18900</td>\n",
       "      <td>18.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean_dt_TV1_TV4</th>\n",
       "      <td>18900</td>\n",
       "      <td>18.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min_dt_TV1_TV4</th>\n",
       "      <td>18900</td>\n",
       "      <td>18.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean_dt_TV1</th>\n",
       "      <td>3191</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min_dt_TV1</th>\n",
       "      <td>3191</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>med_dt_TV1</th>\n",
       "      <td>3191</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>OTR</th>\n",
       "      <td>1257</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_lifetime_cancel</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>MeanNbTradesBySecond</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>MaxNbTradesBySecond</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NbTradeVenueMic</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>type</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90_p_lifetime_cancel</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75_p_lifetime_cancel</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25_p_lifetime_cancel</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>med_lifetime_cancel</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10_p_lifetime_cancel</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean_lifetime_cancel</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min_lifetime_cancel</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_time_two_events</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90_p_time_two_events</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75_p_time_two_events</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25_p_time_two_events</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>med_time_two_events</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10_p_time_two_events</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean_time_two_events</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min_time_two_events</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>OCR</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NbSecondWithAtLeatOneTrade</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                            nbre de NaN  % des NaN\n",
       "OMR                               35710       34.0\n",
       "med_dt_TV1_TV3                    18974       18.0\n",
       "mean_dt_TV1_TV3                   18974       18.0\n",
       "min_dt_TV1_TV3                    18974       18.0\n",
       "med_dt_TV1_TV2                    18917       18.0\n",
       "mean_dt_TV1_TV2                   18917       18.0\n",
       "min_dt_TV1_TV2                    18917       18.0\n",
       "med_dt_TV1_TV4                    18900       18.0\n",
       "mean_dt_TV1_TV4                   18900       18.0\n",
       "min_dt_TV1_TV4                    18900       18.0\n",
       "mean_dt_TV1                        3191        3.0\n",
       "min_dt_TV1                         3191        3.0\n",
       "med_dt_TV1                         3191        3.0\n",
       "OTR                                1257        1.0\n",
       "max_lifetime_cancel                   0        0.0\n",
       "MeanNbTradesBySecond                  0        0.0\n",
       "MaxNbTradesBySecond                   0        0.0\n",
       "NbTradeVenueMic                       0        0.0\n",
       "type                                  0        0.0\n",
       "90_p_lifetime_cancel                  0        0.0\n",
       "75_p_lifetime_cancel                  0        0.0\n",
       "25_p_lifetime_cancel                  0        0.0\n",
       "med_lifetime_cancel                   0        0.0\n",
       "10_p_lifetime_cancel                  0        0.0\n",
       "mean_lifetime_cancel                  0        0.0\n",
       "min_lifetime_cancel                   0        0.0\n",
       "max_time_two_events                   0        0.0\n",
       "90_p_time_two_events                  0        0.0\n",
       "75_p_time_two_events                  0        0.0\n",
       "25_p_time_two_events                  0        0.0\n",
       "med_time_two_events                   0        0.0\n",
       "10_p_time_two_events                  0        0.0\n",
       "mean_time_two_events                  0        0.0\n",
       "min_time_two_events                   0        0.0\n",
       "OCR                                   0        0.0\n",
       "NbSecondWithAtLeatOneTrade            0        0.0"
      ]
     },
     "execution_count": 573,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# déterminer pour chaque variable le nbre de valeurs manquantes et leur part dans le total des observations ( sur le training set )\n",
    "nb_NaN=df_train.drop(['Trader'],axis=1).isnull().sum().sort_values(ascending=False)\n",
    "df_null=pd.DataFrame({'nbre de NaN' : nb_NaN,'% des NaN':round((nb_NaN/df1.shape[0])*100)})\n",
    "df_null"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d7553660",
   "metadata": {},
   "source": [
    "#### <font color='red'><u>Commentaire</u></font> :  Un peu moins que la moitié des valeurs de la variable OMR sont manquantes mais on va garder cette variable car , à priori ,elle semble importante ( à vérifier dans la partie Feature selection ) . Elle traduit le % d'évènements de type modification de l'ordre qui semble justifier en partie le type du trader . \n",
    "#### --> on va procéder (aussi bien pour la variable OMR que OTR) à un remplacement par la moyenne ."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7c743890",
   "metadata": {},
   "source": [
    "#### - <u>Traiter les valeurs manquantes des variables OMR et OTR</u>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 574,
   "id": "07e6dfc3",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\ASUS\\anaconda3\\lib\\site-packages\\pandas\\core\\frame.py:3636: 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: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self[k1] = value[k2]\n"
     ]
    }
   ],
   "source": [
    "moy_stockés1=df_train[['OMR','OTR']].mean()  #on stocke les moyennes du training set pour les remplacer après dans le testing set\n",
    "df_train[['OMR','OTR']]=df_train[['OMR','OTR']].fillna(moy_stockés1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 575,
   "id": "f491e137",
   "metadata": {},
   "outputs": [],
   "source": [
    "#on remplace par les moyennes du training set dans le testing set\n",
    "df_test[['OMR','OTR']]=df_test[['OMR','OTR']].fillna(moy_stockés1) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 576,
   "id": "ae466025",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Trader                            0\n",
       "type                              0\n",
       "OTR                               0\n",
       "OCR                               0\n",
       "OMR                               0\n",
       "min_time_two_events               0\n",
       "mean_time_two_events              0\n",
       "10_p_time_two_events              0\n",
       "med_time_two_events               0\n",
       "25_p_time_two_events              0\n",
       "75_p_time_two_events              0\n",
       "90_p_time_two_events              0\n",
       "max_time_two_events               0\n",
       "min_lifetime_cancel               0\n",
       "mean_lifetime_cancel              0\n",
       "10_p_lifetime_cancel              0\n",
       "med_lifetime_cancel               0\n",
       "25_p_lifetime_cancel              0\n",
       "75_p_lifetime_cancel              0\n",
       "90_p_lifetime_cancel              0\n",
       "max_lifetime_cancel               0\n",
       "NbTradeVenueMic                   0\n",
       "MaxNbTradesBySecond               0\n",
       "MeanNbTradesBySecond              0\n",
       "min_dt_TV1                     3191\n",
       "mean_dt_TV1                    3191\n",
       "med_dt_TV1                     3191\n",
       "min_dt_TV1_TV2                18917\n",
       "mean_dt_TV1_TV2               18917\n",
       "med_dt_TV1_TV2                18917\n",
       "min_dt_TV1_TV3                18974\n",
       "mean_dt_TV1_TV3               18974\n",
       "med_dt_TV1_TV3                18974\n",
       "min_dt_TV1_TV4                18900\n",
       "mean_dt_TV1_TV4               18900\n",
       "med_dt_TV1_TV4                18900\n",
       "NbSecondWithAtLeatOneTrade        0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 576,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6d1875f2",
   "metadata": {},
   "source": [
    "#### - <u>Traiter les autres valeurs manquantes</u>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ca1fd665",
   "metadata": {},
   "source": [
    "Vu que le temps entre 2 trades du TV1 puis du TV2 , ou du TV1 puis du TV3 , ou du TV1 puis du TV4 traduisent la même information\n",
    "à savoir le temps entre 2 trades de 2 trading venues différentes , on a décidé d'agréger les variables \n",
    "<font color='red'>**[ 'min_dt_TV1_TV2','mean_dt_TV1_TV2', 'med_dt_TV1_TV2', 'min_dt_TV1_TV3','mean_dt_TV1_TV3', 'med_dt_TV1_TV3', 'min_dt_TV1_TV4',\n",
    "'mean_dt_TV1_TV4', 'med_dt_TV1_TV4' ]**</font> de la facon suivante :\n",
    "\n",
    "min_dt_TV1_TV2 / min_dt_TV1_TV3 / min_dt_TV1_TV4 **-->** moyenne : le temps minimum moyen entre 2 trades de 2 venues différentes\n",
    "\n",
    "med_dt_TV1_TV2 / med_dt_TV1_TV3 / med_dt_TV1_TV4 **-->** moyenne : le temps médian moyen entre 2 trades de 2 venues différentes\n",
    "\n",
    "mean_dt_TV1_TV2 / mean_dt_TV1_TV3 / mean_dt_TV1_TV4 **-->** moyenne : la moyenne du temps moyen entre 2 trades de 2 venues différentes\n",
    "\n",
    "*Autre le fait que ces 9 variables traduisent la même information , On les agrège aussi dans une tentative de réduire les valeurs manquantes ( la moyenne de 1 NaN et 2 valeurs non NaN donnent une valeur non NaN ) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 577,
   "id": "ba77883e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                min_dt_TV1_TV2  min_dt_TV1_TV3  min_dt_TV1_TV4\n",
      "min_dt_TV1_TV2        1.000000        0.957061        0.946751\n",
      "min_dt_TV1_TV3        0.957061        1.000000        0.967721\n",
      "min_dt_TV1_TV4        0.946751        0.967721        1.000000                  mean_dt_TV1_TV2  mean_dt_TV1_TV3  mean_dt_TV1_TV4\n",
      "mean_dt_TV1_TV2         1.000000         0.976354         0.968322\n",
      "mean_dt_TV1_TV3         0.976354         1.000000         0.979244\n",
      "mean_dt_TV1_TV4         0.968322         0.979244         1.000000                  mean_dt_TV1_TV2  mean_dt_TV1_TV3  mean_dt_TV1_TV4\n",
      "mean_dt_TV1_TV2         1.000000         0.976354         0.968322\n",
      "mean_dt_TV1_TV3         0.976354         1.000000         0.979244\n",
      "mean_dt_TV1_TV4         0.968322         0.979244         1.000000\n"
     ]
    }
   ],
   "source": [
    "# Pour confirmer le fait que les 2 trades du TV1 puis du TV2 , ou du TV1 puis du TV3 , ou du TV1 puis du TV4 \n",
    "# traduisent la même information : leur corrélation est très forte > 0.9\n",
    "print(df_train[['min_dt_TV1_TV2','min_dt_TV1_TV3','min_dt_TV1_TV4']].corr(),\n",
    "      df_train[['mean_dt_TV1_TV2','mean_dt_TV1_TV3','mean_dt_TV1_TV4']].corr() \n",
    "      , df_train[['mean_dt_TV1_TV2','mean_dt_TV1_TV3','mean_dt_TV1_TV4']].corr() )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 578,
   "id": "0cd1017b",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\ASUS\\anaconda3\\lib\\site-packages\\ipykernel_launcher.py:3: 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: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  This is separate from the ipykernel package so we can avoid doing imports until\n",
      "C:\\Users\\ASUS\\anaconda3\\lib\\site-packages\\ipykernel_launcher.py:4: 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: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  after removing the cwd from sys.path.\n",
      "C:\\Users\\ASUS\\anaconda3\\lib\\site-packages\\ipykernel_launcher.py:5: 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: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  \"\"\"\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "moy_mean_dt_TVi_TVj           18739\n",
       "moy_med_dt_TVi_TVj            18739\n",
       "moy_min_dt_TVi_TVj            18739\n",
       "med_dt_TV1                     3191\n",
       "mean_dt_TV1                    3191\n",
       "min_dt_TV1                     3191\n",
       "med_lifetime_cancel               0\n",
       "NbSecondWithAtLeatOneTrade        0\n",
       "MeanNbTradesBySecond              0\n",
       "MaxNbTradesBySecond               0\n",
       "NbTradeVenueMic                   0\n",
       "max_lifetime_cancel               0\n",
       "90_p_lifetime_cancel              0\n",
       "75_p_lifetime_cancel              0\n",
       "25_p_lifetime_cancel              0\n",
       "Trader                            0\n",
       "type                              0\n",
       "mean_lifetime_cancel              0\n",
       "min_lifetime_cancel               0\n",
       "max_time_two_events               0\n",
       "90_p_time_two_events              0\n",
       "75_p_time_two_events              0\n",
       "25_p_time_two_events              0\n",
       "med_time_two_events               0\n",
       "10_p_time_two_events              0\n",
       "mean_time_two_events              0\n",
       "min_time_two_events               0\n",
       "OMR                               0\n",
       "OCR                               0\n",
       "OTR                               0\n",
       "10_p_lifetime_cancel              0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 578,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# pour éviter de refaire le travail 2 fois , on définit une fct° . On appelera cette fct° sur le training et le testing set \n",
    "def moy_tvi_tvj(df):\n",
    "    df['moy_min_dt_TVi_TVj']=df[['min_dt_TV1_TV2','min_dt_TV1_TV3','min_dt_TV1_TV4']].mean(axis=1)\n",
    "    df['moy_med_dt_TVi_TVj']=df[['med_dt_TV1_TV2','med_dt_TV1_TV3','med_dt_TV1_TV4']].mean(axis=1)\n",
    "    df['moy_mean_dt_TVi_TVj']=df[['mean_dt_TV1_TV2','mean_dt_TV1_TV3','mean_dt_TV1_TV4']].mean(axis=1)\n",
    "    df=df.drop(['min_dt_TV1_TV2',\n",
    "           'mean_dt_TV1_TV2', 'med_dt_TV1_TV2', 'min_dt_TV1_TV3',\n",
    "           'mean_dt_TV1_TV3', 'med_dt_TV1_TV3', 'min_dt_TV1_TV4',\n",
    "           'mean_dt_TV1_TV4', 'med_dt_TV1_TV4'],axis=1)\n",
    "    return df\n",
    "\n",
    "df_train=moy_tvi_tvj(df_train)\n",
    "df_test=moy_tvi_tvj(df_test)\n",
    "df_train.isnull().sum().sort_values(ascending=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5f172dd7",
   "metadata": {},
   "source": [
    "Même avec l'aggrégation de ces 9 variables , les valeurs manquantes sont encore élevés (18739).  Les 3 variables dt_TVi_TVj \n",
    "traduisent une information importante (<font color='red'>à priori</font> ) : Les traders HFT tendent à effectuer des trades dans des différentes venues\n",
    "dans un laps de temps plus court ( **la simultanéité ou quasi-simultanéité de 2 trades dans 2 places financières différentes\n",
    "est une caractéristique probable des traders HFT** ) . \n",
    "Donc , on va garder les 3 variables dt_TVi_TVj ainsi que les variables : min_dt_TV1', 'mean_dt_TV1', 'med_dt_TV1'. On va \n",
    "procéder de la même manière pour le remplacement des valeurs manquantes . "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 579,
   "id": "97bec007",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['moy_min_dt_TVi_TVj',\n",
       " 'moy_med_dt_TVi_TVj',\n",
       " 'moy_mean_dt_TVi_TVj',\n",
       " 'min_dt_TV1',\n",
       " 'mean_dt_TV1',\n",
       " 'med_dt_TV1']"
      ]
     },
     "execution_count": 579,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lst=list(df_train.columns)\n",
    "lst_names=lst[-3:]+lst[-7:-4]\n",
    "lst_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 580,
   "id": "c7f9a38c",
   "metadata": {},
   "outputs": [],
   "source": [
    "moy_stockés2=df_train[lst_names].mean()  #on stocke les moyennes du training set pour les remplacer après dans le testing set\n",
    "df_train[lst_names]=df_train[lst_names].fillna(moy_stockés2)\n",
    "df_test[lst_names]=df_test[lst_names].fillna(moy_stockés2) #on remplace par les moyennes du training set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 581,
   "id": "dfec830f",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Trader                        0\n",
       "type                          0\n",
       "OTR                           0\n",
       "OCR                           0\n",
       "OMR                           0\n",
       "min_time_two_events           0\n",
       "mean_time_two_events          0\n",
       "10_p_time_two_events          0\n",
       "med_time_two_events           0\n",
       "25_p_time_two_events          0\n",
       "75_p_time_two_events          0\n",
       "90_p_time_two_events          0\n",
       "max_time_two_events           0\n",
       "min_lifetime_cancel           0\n",
       "mean_lifetime_cancel          0\n",
       "10_p_lifetime_cancel          0\n",
       "med_lifetime_cancel           0\n",
       "25_p_lifetime_cancel          0\n",
       "75_p_lifetime_cancel          0\n",
       "90_p_lifetime_cancel          0\n",
       "max_lifetime_cancel           0\n",
       "NbTradeVenueMic               0\n",
       "MaxNbTradesBySecond           0\n",
       "MeanNbTradesBySecond          0\n",
       "min_dt_TV1                    0\n",
       "mean_dt_TV1                   0\n",
       "med_dt_TV1                    0\n",
       "NbSecondWithAtLeatOneTrade    0\n",
       "moy_min_dt_TVi_TVj            0\n",
       "moy_med_dt_TVi_TVj            0\n",
       "moy_mean_dt_TVi_TVj           0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 581,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "57b41114",
   "metadata": {},
   "source": [
    "#### --> on n'a plus de valeurs manquantes"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fab74f87",
   "metadata": {},
   "source": [
    "### <u>4- aggrégation de la base de données  <u>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 582,
   "id": "fb5dbc5d",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Aggrégation de la base de données : moyenne des observations par trader\n",
    "df_train=df_train.groupby(['Trader']).mean().reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 583,
   "id": "a9d77890",
   "metadata": {
    "scrolled": 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>Trader</th>\n",
       "      <th>type</th>\n",
       "      <th>OTR</th>\n",
       "      <th>OCR</th>\n",
       "      <th>OMR</th>\n",
       "      <th>min_time_two_events</th>\n",
       "      <th>mean_time_two_events</th>\n",
       "      <th>10_p_time_two_events</th>\n",
       "      <th>med_time_two_events</th>\n",
       "      <th>25_p_time_two_events</th>\n",
       "      <th>...</th>\n",
       "      <th>NbTradeVenueMic</th>\n",
       "      <th>MaxNbTradesBySecond</th>\n",
       "      <th>MeanNbTradesBySecond</th>\n",
       "      <th>min_dt_TV1</th>\n",
       "      <th>mean_dt_TV1</th>\n",
       "      <th>med_dt_TV1</th>\n",
       "      <th>NbSecondWithAtLeatOneTrade</th>\n",
       "      <th>moy_min_dt_TVi_TVj</th>\n",
       "      <th>moy_med_dt_TVi_TVj</th>\n",
       "      <th>moy_mean_dt_TVi_TVj</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Trader_285</td>\n",
       "      <td>MIX</td>\n",
       "      <td>7.807554</td>\n",
       "      <td>34.600000</td>\n",
       "      <td>163.730280</td>\n",
       "      <td>8.951034</td>\n",
       "      <td>864.050787</td>\n",
       "      <td>8.951034</td>\n",
       "      <td>139.254886</td>\n",
       "      <td>8.951054</td>\n",
       "      <td>...</td>\n",
       "      <td>9.200000</td>\n",
       "      <td>17.400000</td>\n",
       "      <td>5.119288</td>\n",
       "      <td>87.691021</td>\n",
       "      <td>226.333150</td>\n",
       "      <td>130.584267</td>\n",
       "      <td>39.000000</td>\n",
       "      <td>139.513929</td>\n",
       "      <td>196.813139</td>\n",
       "      <td>430.459508</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Trader_114</td>\n",
       "      <td>NON HFT</td>\n",
       "      <td>5.291763</td>\n",
       "      <td>26.348230</td>\n",
       "      <td>163.796670</td>\n",
       "      <td>281.143544</td>\n",
       "      <td>7327.222269</td>\n",
       "      <td>285.624589</td>\n",
       "      <td>1951.956669</td>\n",
       "      <td>497.031552</td>\n",
       "      <td>...</td>\n",
       "      <td>1.948348</td>\n",
       "      <td>14.722523</td>\n",
       "      <td>1.476684</td>\n",
       "      <td>118.461499</td>\n",
       "      <td>743.325773</td>\n",
       "      <td>400.064941</td>\n",
       "      <td>104.612012</td>\n",
       "      <td>250.939310</td>\n",
       "      <td>489.069498</td>\n",
       "      <td>796.776108</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Trader_110</td>\n",
       "      <td>NON HFT</td>\n",
       "      <td>3.666667</td>\n",
       "      <td>9.142857</td>\n",
       "      <td>163.730280</td>\n",
       "      <td>1725.812961</td>\n",
       "      <td>14833.008158</td>\n",
       "      <td>1725.812961</td>\n",
       "      <td>3325.696247</td>\n",
       "      <td>1725.812961</td>\n",
       "      <td>...</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.571429</td>\n",
       "      <td>2.428571</td>\n",
       "      <td>6455.785804</td>\n",
       "      <td>7946.697951</td>\n",
       "      <td>6578.318506</td>\n",
       "      <td>1.571429</td>\n",
       "      <td>311.478322</td>\n",
       "      <td>403.644681</td>\n",
       "      <td>613.857998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Trader_57</td>\n",
       "      <td>NON HFT</td>\n",
       "      <td>3.261453</td>\n",
       "      <td>11.372549</td>\n",
       "      <td>94.474854</td>\n",
       "      <td>341.188804</td>\n",
       "      <td>4456.655164</td>\n",
       "      <td>343.341275</td>\n",
       "      <td>601.566444</td>\n",
       "      <td>348.091122</td>\n",
       "      <td>...</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>2.705882</td>\n",
       "      <td>1.899160</td>\n",
       "      <td>3193.790743</td>\n",
       "      <td>6042.941310</td>\n",
       "      <td>4087.296235</td>\n",
       "      <td>3.470588</td>\n",
       "      <td>311.478322</td>\n",
       "      <td>403.644681</td>\n",
       "      <td>613.857998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Trader_128</td>\n",
       "      <td>NON HFT</td>\n",
       "      <td>9.888919</td>\n",
       "      <td>12.624458</td>\n",
       "      <td>163.730280</td>\n",
       "      <td>3669.775350</td>\n",
       "      <td>9966.792043</td>\n",
       "      <td>3669.775354</td>\n",
       "      <td>5112.676497</td>\n",
       "      <td>3705.061185</td>\n",
       "      <td>...</td>\n",
       "      <td>1.419753</td>\n",
       "      <td>5.769547</td>\n",
       "      <td>1.862654</td>\n",
       "      <td>1173.328002</td>\n",
       "      <td>2680.564504</td>\n",
       "      <td>1545.688415</td>\n",
       "      <td>13.646091</td>\n",
       "      <td>1114.433802</td>\n",
       "      <td>1195.094042</td>\n",
       "      <td>1609.667081</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 31 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       Trader     type       OTR        OCR         OMR  min_time_two_events  \\\n",
       "0  Trader_285      MIX  7.807554  34.600000  163.730280             8.951034   \n",
       "1  Trader_114  NON HFT  5.291763  26.348230  163.796670           281.143544   \n",
       "2  Trader_110  NON HFT  3.666667   9.142857  163.730280          1725.812961   \n",
       "3   Trader_57  NON HFT  3.261453  11.372549   94.474854           341.188804   \n",
       "4  Trader_128  NON HFT  9.888919  12.624458  163.730280          3669.775350   \n",
       "\n",
       "   mean_time_two_events  10_p_time_two_events  med_time_two_events  \\\n",
       "0            864.050787              8.951034           139.254886   \n",
       "1           7327.222269            285.624589          1951.956669   \n",
       "2          14833.008158           1725.812961          3325.696247   \n",
       "3           4456.655164            343.341275           601.566444   \n",
       "4           9966.792043           3669.775354          5112.676497   \n",
       "\n",
       "   25_p_time_two_events  ...  NbTradeVenueMic  MaxNbTradesBySecond  \\\n",
       "0              8.951054  ...         9.200000            17.400000   \n",
       "1            497.031552  ...         1.948348            14.722523   \n",
       "2           1725.812961  ...         1.000000             3.571429   \n",
       "3            348.091122  ...         1.000000             2.705882   \n",
       "4           3705.061185  ...         1.419753             5.769547   \n",
       "\n",
       "   MeanNbTradesBySecond   min_dt_TV1  mean_dt_TV1   med_dt_TV1  \\\n",
       "0              5.119288    87.691021   226.333150   130.584267   \n",
       "1              1.476684   118.461499   743.325773   400.064941   \n",
       "2              2.428571  6455.785804  7946.697951  6578.318506   \n",
       "3              1.899160  3193.790743  6042.941310  4087.296235   \n",
       "4              1.862654  1173.328002  2680.564504  1545.688415   \n",
       "\n",
       "   NbSecondWithAtLeatOneTrade  moy_min_dt_TVi_TVj  moy_med_dt_TVi_TVj  \\\n",
       "0                   39.000000          139.513929          196.813139   \n",
       "1                  104.612012          250.939310          489.069498   \n",
       "2                    1.571429          311.478322          403.644681   \n",
       "3                    3.470588          311.478322          403.644681   \n",
       "4                   13.646091         1114.433802         1195.094042   \n",
       "\n",
       "   moy_mean_dt_TVi_TVj  \n",
       "0           430.459508  \n",
       "1           796.776108  \n",
       "2           613.857998  \n",
       "3           613.857998  \n",
       "4          1609.667081  \n",
       "\n",
       "[5 rows x 31 columns]"
      ]
     },
     "execution_count": 583,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train=pd.merge(df_y,df_train,on='Trader')\n",
    "df_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 584,
   "id": "92aeccea",
   "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>Trader</th>\n",
       "      <th>type</th>\n",
       "      <th>OTR</th>\n",
       "      <th>OCR</th>\n",
       "      <th>OMR</th>\n",
       "      <th>min_time_two_events</th>\n",
       "      <th>mean_time_two_events</th>\n",
       "      <th>10_p_time_two_events</th>\n",
       "      <th>med_time_two_events</th>\n",
       "      <th>25_p_time_two_events</th>\n",
       "      <th>...</th>\n",
       "      <th>NbTradeVenueMic</th>\n",
       "      <th>MaxNbTradesBySecond</th>\n",
       "      <th>MeanNbTradesBySecond</th>\n",
       "      <th>min_dt_TV1</th>\n",
       "      <th>mean_dt_TV1</th>\n",
       "      <th>med_dt_TV1</th>\n",
       "      <th>NbSecondWithAtLeatOneTrade</th>\n",
       "      <th>moy_min_dt_TVi_TVj</th>\n",
       "      <th>moy_med_dt_TVi_TVj</th>\n",
       "      <th>moy_mean_dt_TVi_TVj</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Trader_285</td>\n",
       "      <td>MIX</td>\n",
       "      <td>25.853940</td>\n",
       "      <td>7.250000</td>\n",
       "      <td>163.730280</td>\n",
       "      <td>15945.150000</td>\n",
       "      <td>15988.277334</td>\n",
       "      <td>15945.150005</td>\n",
       "      <td>15945.150112</td>\n",
       "      <td>15945.150019</td>\n",
       "      <td>...</td>\n",
       "      <td>8.000000</td>\n",
       "      <td>11.000000</td>\n",
       "      <td>4.109375</td>\n",
       "      <td>0.017000</td>\n",
       "      <td>64.014726</td>\n",
       "      <td>0.021500</td>\n",
       "      <td>35.000000</td>\n",
       "      <td>745.809667</td>\n",
       "      <td>745.864500</td>\n",
       "      <td>755.882771</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Trader_114</td>\n",
       "      <td>NON HFT</td>\n",
       "      <td>5.330605</td>\n",
       "      <td>26.173368</td>\n",
       "      <td>163.730280</td>\n",
       "      <td>163.240074</td>\n",
       "      <td>7346.201169</td>\n",
       "      <td>165.405602</td>\n",
       "      <td>1885.117576</td>\n",
       "      <td>363.844437</td>\n",
       "      <td>...</td>\n",
       "      <td>1.969203</td>\n",
       "      <td>16.317029</td>\n",
       "      <td>1.479432</td>\n",
       "      <td>150.485179</td>\n",
       "      <td>705.924973</td>\n",
       "      <td>395.613114</td>\n",
       "      <td>107.835145</td>\n",
       "      <td>257.875933</td>\n",
       "      <td>480.769582</td>\n",
       "      <td>767.662607</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Trader_110</td>\n",
       "      <td>NON HFT</td>\n",
       "      <td>2.400000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>163.730280</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>15262.259766</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>2.500000</td>\n",
       "      <td>604.070402</td>\n",
       "      <td>604.070402</td>\n",
       "      <td>604.070402</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>311.478322</td>\n",
       "      <td>403.644681</td>\n",
       "      <td>613.857998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Trader_57</td>\n",
       "      <td>NON HFT</td>\n",
       "      <td>2.396578</td>\n",
       "      <td>18.055556</td>\n",
       "      <td>99.989045</td>\n",
       "      <td>0.281090</td>\n",
       "      <td>5183.910167</td>\n",
       "      <td>0.281090</td>\n",
       "      <td>711.692024</td>\n",
       "      <td>0.281333</td>\n",
       "      <td>...</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>4.111111</td>\n",
       "      <td>2.013533</td>\n",
       "      <td>1372.979357</td>\n",
       "      <td>3086.425393</td>\n",
       "      <td>1960.297629</td>\n",
       "      <td>4.888889</td>\n",
       "      <td>311.478322</td>\n",
       "      <td>403.644681</td>\n",
       "      <td>613.857998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Trader_128</td>\n",
       "      <td>NON HFT</td>\n",
       "      <td>9.973350</td>\n",
       "      <td>11.503487</td>\n",
       "      <td>163.730280</td>\n",
       "      <td>3023.794192</td>\n",
       "      <td>8835.333871</td>\n",
       "      <td>3023.794193</td>\n",
       "      <td>5269.297803</td>\n",
       "      <td>3211.912981</td>\n",
       "      <td>...</td>\n",
       "      <td>1.464286</td>\n",
       "      <td>5.571429</td>\n",
       "      <td>1.603555</td>\n",
       "      <td>1371.379622</td>\n",
       "      <td>2991.888247</td>\n",
       "      <td>1791.915847</td>\n",
       "      <td>17.869048</td>\n",
       "      <td>1112.806234</td>\n",
       "      <td>1186.309211</td>\n",
       "      <td>1497.133224</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 31 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       Trader     type        OTR        OCR         OMR  min_time_two_events  \\\n",
       "0  Trader_285      MIX  25.853940   7.250000  163.730280         15945.150000   \n",
       "1  Trader_114  NON HFT   5.330605  26.173368  163.730280           163.240074   \n",
       "2  Trader_110  NON HFT   2.400000   6.000000  163.730280             0.000000   \n",
       "3   Trader_57  NON HFT   2.396578  18.055556   99.989045             0.281090   \n",
       "4  Trader_128  NON HFT   9.973350  11.503487  163.730280          3023.794192   \n",
       "\n",
       "   mean_time_two_events  10_p_time_two_events  med_time_two_events  \\\n",
       "0          15988.277334          15945.150005         15945.150112   \n",
       "1           7346.201169            165.405602          1885.117576   \n",
       "2          15262.259766              0.000000             0.000000   \n",
       "3           5183.910167              0.281090           711.692024   \n",
       "4           8835.333871           3023.794193          5269.297803   \n",
       "\n",
       "   25_p_time_two_events  ...  NbTradeVenueMic  MaxNbTradesBySecond  \\\n",
       "0          15945.150019  ...         8.000000            11.000000   \n",
       "1            363.844437  ...         1.969203            16.317029   \n",
       "2              0.000000  ...         1.000000             4.000000   \n",
       "3              0.281333  ...         1.000000             4.111111   \n",
       "4           3211.912981  ...         1.464286             5.571429   \n",
       "\n",
       "   MeanNbTradesBySecond   min_dt_TV1  mean_dt_TV1   med_dt_TV1  \\\n",
       "0              4.109375     0.017000    64.014726     0.021500   \n",
       "1              1.479432   150.485179   705.924973   395.613114   \n",
       "2              2.500000   604.070402   604.070402   604.070402   \n",
       "3              2.013533  1372.979357  3086.425393  1960.297629   \n",
       "4              1.603555  1371.379622  2991.888247  1791.915847   \n",
       "\n",
       "   NbSecondWithAtLeatOneTrade  moy_min_dt_TVi_TVj  moy_med_dt_TVi_TVj  \\\n",
       "0                   35.000000          745.809667          745.864500   \n",
       "1                  107.835145          257.875933          480.769582   \n",
       "2                    2.000000          311.478322          403.644681   \n",
       "3                    4.888889          311.478322          403.644681   \n",
       "4                   17.869048         1112.806234         1186.309211   \n",
       "\n",
       "   moy_mean_dt_TVi_TVj  \n",
       "0           755.882771  \n",
       "1           767.662607  \n",
       "2           613.857998  \n",
       "3           613.857998  \n",
       "4          1497.133224  \n",
       "\n",
       "[5 rows x 31 columns]"
      ]
     },
     "execution_count": 584,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test=df_test.groupby(['Trader']).mean().reset_index()\n",
    "df_test=pd.merge(df_y,df_test,on='Trader')\n",
    "df_test.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 585,
   "id": "f9da9414",
   "metadata": {},
   "outputs": [],
   "source": [
    "#séparer la colonne du target de la matrice des variables explicatives + encoding de la colonne du target \n",
    "y_train=df_train['type'].replace({'HFT':0,'NON HFT':1,'MIX':2})\n",
    "x_train=df_train.drop(['type','Trader'],axis=1) \n",
    "y_test=df_test['type'].replace({'HFT':0,'NON HFT':1,'MIX':2})\n",
    "x_test=df_test.drop(['type','Trader'],axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1c05c371",
   "metadata": {},
   "source": [
    " ### <u>5- Data exploration"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 586,
   "id": "4d62dc31",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:title={'center':'Distribution du Target'}, ylabel='% des traders'>"
      ]
     },
     "execution_count": 586,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAElCAYAAADjk4nIAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8rg+JYAAAACXBIWXMAAAsTAAALEwEAmpwYAAAagElEQVR4nO3deZhdVZ3u8e9LiIxhCAQIQogMzSN6UekoeLWvAyCjBlpBBDUKijyKojghIqItSve1ESeQQZoIXgRvQ4O0U0BpZzAgXFG0o8gcSABDQpApvPePvUqORaXqJKl9TlXW+3mees7Zw9n7Vzmp9+yz9tpryzYREVGPNfpdQERE9FaCPyKiMgn+iIjKJPgjIiqT4I+IqEyCPyKiMgn+aJ2kr0j62Chta5qkhyRNKNNXS3rbaGy7bO87kmaN1vaG2c9bJP2k7f1EDCXBH6tE0q2S/iJpiaRFkn4m6ShJf/2/Zfso2//U5bb2GG4d27fbXt/2slGo/SRJFwza/j62Z6/qtttSPkQfKj+PSXq8Y/o7PaxjxPcqxq4Ef4yGV9ueBGwDnAJ8GPjqaO9E0pqjvc3xpnyIrm97feDTwEUD07b36XY7A9+Yok4J/hg1th+0fTnwemCWpOcCSDpP0qfK800lXVG+HTwg6ceS1pB0PjAN+FY5ev2QpOmSLOkISbcDP+iY1/khsJ2kayU9KOkySZPLvl4u6c7OGgeOVCXtDRwPvL7s78ay/K9NR6WuEyTdJmmBpK9J2rAsG6hjlqTbJd0n6aPL+7eRtImkyyUtlnQtsF3Hsqf9TivThCXpm5LuKf8OP5L0nI5l50k6Q9K3JS0FXiFpF0m/Kt/WvinpooH3qbxmf0k3dHyT27nMf9p7tSJ1Rv8l+GPU2b4WuBP4hyEWv78smwJsThO+tv0m4Haabw/r2/6Xjte8DHg2sNdydvlm4HBgS+AJ4Atd1Phd/vaI+XlDrPaW8vMKYFtgfeBLg9Z5KbAjsDtwoqRnL2eXXwYeAaaWWg8fqcaV8B1gB2Az4Hrg64OWHwqcDEwCrgUuBc4DJgMXAgcOrChpF+Bc4B3AJsCZwOWS1hrhvYpxIMEfbbmbJlAGe5wm/Lax/bjtH3vkAaNOsr3U9l+Ws/x82zfZXgp8DDh4lJoyDgNOtX2L7YeAjwCHDPq28Qnbf7F9I3Aj8LQPkFLLa4ETy+9xEzDq5xFsn2t7ie1HgZOA5w18Qykus/1T208CzwfWBL5Q3odLaD4MBrwdONP2NbaXlfMejwK7jXbd0XsJ/mjLM4EHhpj/v4E/AN+XdIuk47rY1h0rsPw2YCKwaVdVDm/Lsr3Oba9J801lwD0dzx+m+VYw2JTyusF1jhpJEySdIumPkhYDt5ZFnf8OnfvfErhr0Idu5/JtgPeXZp5FkhYBW5fXxTiX4I9RJ+mFNMH/tO6K5Yj0/ba3BV4NHCtp94HFy9nkSN8Itu54Po3mW8V9wFJg3Y66JtCEcLfbvZsmADu3/QRw7wivG2xhed3gOgcsLY/rdszbYgX3cSgwE9gD2BCYXuarY53O33c+8ExJncs767sDONn2Rh0/69q+cIhtxTiT4I9RI2kDSfsD3wAusP3rIdbZX9L2JXAWA8vKDzSBuu1K7PqNknaStC7wSeD/lu6e/w2sLWk/SROBE4C1Ol53LzBdHV1PB7kQeJ+kZ0nq7EXzxIoUV2q5BDhJ0rqSdgJmdSxfCNxVfo8Jkg6n4+RvlybRNMXcT/MB8ukR1v85zb/70ZLWlDQTeFHH8rOBoyTtqsZ65d9xUlm+su9VjAEJ/hgN35K0hOYo8aPAqcBbl7PuDsCVwEM04XO67avLss8AJ5SmhQ+swP7PpzlJeQ+wNvAeaHoZAe8EzqEJ1qU0J5YHfLM83i/p+iG2e27Z9o+AP9GcnH33CtTV6WiaZqB7Sq3/Nmj524EP0gT3c4CfreD2v0bTfHQX8FvgF8OtbPsx4B+BI4BFwBuBK2g+PLA9t9T0JeDPNM1zb+nYxMq+VzEGKDdiiQgASdcAX7E9+EMpVjM54o+olKSXSdqiNPXMAnYGvtvvuqJ91V8JGVGxHYGLaZqg/gi8zvb8/pYUvZCmnoiIyqSpJyKiMuOiqWfTTTf19OnT+11GRMS4ct11191ne8rg+eMi+KdPn87cuXP7XUZExLgiacgrxNPUExFRmQR/RERlEvwREZVJ8EdEVCbBHxFRmQR/RERlEvwREZVJ8EdEVCbBHxFRmXFx5W7PnXRSvyto1+r++0XEsHLEHxFRmQR/RERlEvwREZVJ8EdEVCbBHxFRmQR/RERlEvwREZVJ8EdEVCbBHxFRmQR/RERlEvwREZVpdaweSbcCS4BlwBO2Z0iaDFwETAduBQ62/ec264iIiKf04oj/Fbafb3tGmT4OuMr2DsBVZToiInqkH009M4HZ5fls4IA+1BARUa22g9/A9yVdJ+nIMm9z2/MByuNmLdcQEREd2h6P/yW275a0GTBH0u+6fWH5oDgSYNq0aW3VFxFRnVaP+G3fXR4XAJcCLwLulTQVoDwuWM5rz7I9w/aMKVOmtFlmRERVWgt+SetJmjTwHHgVcBNwOTCrrDYLuKytGiIi4unabOrZHLhU0sB+/o/t70r6JXCxpCOA24GDWqwhIiIGaS34bd8CPG+I+fcDu7e134iIGF6u3I2IqEyCPyKiMgn+iIjKJPgjIiqT4I+IqEyCPyKiMgn+iIjKJPgjIiqT4I+IqEyCPyKiMgn+iIjKJPgjIiqT4I+IqEyCPyKiMgn+iIjKJPgjIiqT4I+IqEyCPyKiMgn+iIjKJPgjIiqT4I+IqEyCPyKiMgn+iIjKJPgjIiqT4I+IqEyCPyKiMgn+iIjKJPgjIiqT4I+IqEyCPyKiMq0Hv6QJkn4l6YoyPVnSHEnzyuPGbdcQERFP6cUR/zHAzR3TxwFX2d4BuKpMR0REj7Qa/JK2AvYDzumYPROYXZ7PBg5os4aIiPhbbR/xnwZ8CHiyY97mtucDlMfNhnqhpCMlzZU0d+HChS2XGRFRj9aCX9L+wALb163M622fZXuG7RlTpkwZ5eoiIuq1ZovbfgnwGkn7AmsDG0i6ALhX0lTb8yVNBRa0WENERAzS2hG/7Y/Y3sr2dOAQ4Ae23whcDswqq80CLmurhoiIeLp+9OM/BdhT0jxgzzIdERE90mZTz1/Zvhq4ujy/H9i9F/uNiIiny5W7ERGVSfBHRFQmwR8RUZkEf0REZRL8ERGVSfBHRFRmxOCXdIykDdT4qqTrJb2qF8VFRMTo6+aI/3Dbi4FXAVOAt5KLriIixq1ugl/lcV/g32zf2DEvIiLGmW6C/zpJ36cJ/u9JmsTfDrMcERHjyLBDNkgScCJNE88tth+WtAlNc09ERIxDwwa/bUv6D9t/3zHvfuD+1iuLiIhWdNPU8wtJL2y9koiI6IluRud8BXCUpFuBpTQndm175zYLi4iIdnQT/Pu0XkVERPTMiE09tm8DtgZeWZ4/3M3rIiJibOrmyt2PAx8GPlJmTQQuaLOoiIhoTzdH7gcCr6Fp38f23cCkNouKiIj2dBP8j9k2YABJ67VbUkREtKmb4L9Y0pnARpLeDlwJnN1uWRER0ZYRe/XY/qykPYHFwI7AibbntF5ZRES0opvunJSgT9hHRKwGlhv8kpZQ2vWHYnuDViqKiIhWLTf4bU8CkPRJ4B7gfJqrdg8jvXoiIsatbk7u7mX7dNtLbC+2fQbw2rYLi4iIdnQT/MskHSZpgqQ1JB0GLGu7sIiIaEc3wX8ocDBwb/k5qMyLiIhxqJvunLcCM9svJSIiemHE4Je0NnAE8Bxg7YH5tg9vsa6IiGhJN0095wNbAHsB/wVsBSwZ6UWS1pZ0raQbJf1G0ifK/MmS5kiaVx43XpVfICIiVkw3wb+97Y8BS23PBvYD/kcXr3uUZijn5wHPB/aWtBtwHHCV7R2Aq8p0RET0SDfB/3h5XCTpucCGwPSRXuTGQ2VyYvkxzfmC2WX+bOCAFag3IiJWUTfBf1ZpjjkBuBz4LfDP3Wy8dAG9AVgAzLF9DbC57fkA5XGz5bz2SElzJc1duHBhN7uLiIguDHtyV9IawGLbfwZ+BGy7Ihu3vQx4vqSNgEvLN4ZuX3sWcBbAjBkzljt0RERErJhhj/htPwkcvao7sb0IuBrYG7hX0lSA8rhgVbcfERHd66apZ46kD0jauvTImSxp8kgvkjSlHOkjaR1gD+B3NM1Fs8pqs4DLVq70iIhYGd0MyzzQX/9dHfPMyM0+U4HZkibQfMBcbPsKST+nubnLEcDtNFcCR0REj3QT/M+2/UjnjHJR17Bs/z/gBUPMvx/YvesKIyJiVHXT1POzLudFRMQ4MNyNWLYAngmsI+kFNGPxA2wArNuD2iJW3Ekn9buCdq3uv1/0xHBNPXsBb6EZouFfeSr4FwPHt1tWRES0Zbg7cM2mOTn7Wtv/3sOaIiKiRSO28Sf0IyJWL92c3I2IiNVIgj8iojIjBr+kgyRNKs9PkHSJpF3aLy0iItrQzRH/x2wvkfRSmp4+s4Ez2i0rIiLa0k3wLyuP+wFn2L4MeEZ7JUVERJu6Cf67JJ0JHAx8W9JaXb4uIiLGoG4C/GDge8DeZXjlycAH2ywqIiLa000//odpxsx/aZn1BDCvzaIiIqI93fTq+TjwYeAjZdZE4II2i4qIiPZ009RzIPAaYCmA7buBSW0WFRER7ekm+B+zbZqbryBpvXZLioiINnUT/BeXXj0bSXo7cCVwdrtlRUREW0a8A5ftz0rak2Y45h2BE23Pab2yiIhoRTe3XqQEfcI+ImI1MNwduJZQ2vWHYnuDViqKiIhWDXcjloGB2T4J3AOcT3MXrsNIr56IiHGrm5O7e9k+3fYS24ttnwG8tu3CIiKiHV0N0ibpMEkTJK0h6TCeGrgtIiLGmW6C/1Ca8XruLT8HlXkRETEOddOd81ZgZvulREREL2R45YiIyiT4IyIqk+CPiKhM18EvaTdJP5D0U0kHtFhTRES0aLnBL2mLQbOOpRmeeW/gn0basKStJf1Q0s2SfiPpmDJ/sqQ5kuaVx41X5ReIiIgVM9wR/1ckfUzS2mV6EU03ztfTDNg2kieA99t+NrAb8C5JOwHHAVfZ3gG4qkxHRESPLDf4bR8A3ABcIelNwHuBJ4F1gQNG2rDt+bavL8+XADcDz6TpGjq7rDa7m21FRMToGbaN3/a3gL2AjYBLgN/b/oLthSuyE0nTgRcA1wCb255ftj8f2Gw5rzlS0lxJcxcuXKHdRUTEMIZr43+NpJ8APwBuAg4BDpR0oaTtut2BpPWBfwfea7ubJiIAbJ9le4btGVOmTOn2ZRERMYLhrtz9FPBiYB3g27ZfBBwraQfgZJoPgmFJmkgT+l+3fUmZfa+kqbbnS5oKLFil3yAiIlbIcE09D9KE+yF0hLPteba7CX0BXwVutn1qx6LLgVnl+SzgshUtOiIiVt5wwX8gzYncJ1i5QdleArwJeKWkG8rPvsApwJ6S5gF7lumIiOiR4W7Ech/wxZXdsO2f0Ny4ZSi7r+x2IyJi1WTIhoiIyiT4IyIqk+CPiKjMiDdiiYjomZNO6ncF7Rojv1+O+CMiKpPgj4ioTII/IqIyCf6IiMok+CMiKpPgj4ioTII/IqIyCf6IiMok+CMiKpPgj4ioTII/IqIyCf6IiMok+CMiKpPgj4ioTII/IqIyCf6IiMok+CMiKpPgj4ioTII/IqIyCf6IiMok+CMiKpPgj4ioTII/IqIyCf6IiMok+CMiKtNa8Es6V9ICSTd1zJssaY6keeVx47b2HxERQ2vziP88YO9B844DrrK9A3BVmY6IiB5qLfht/wh4YNDsmcDs8nw2cEBb+4+IiKH1uo1/c9vzAcrjZj3ef0RE9cbsyV1JR0qaK2nuwoUL+11ORMRqo9fBf6+kqQDlccHyVrR9lu0ZtmdMmTKlZwVGRKzueh38lwOzyvNZwGU93n9ERPXa7M55IfBzYEdJd0o6AjgF2FPSPGDPMh0RET20Zlsbtv2G5Szava19RkTEyMbsyd2IiGhHgj8iojIJ/oiIyiT4IyIqk+CPiKhMgj8iojIJ/oiIyiT4IyIqk+CPiKhMgj8iojIJ/oiIyiT4IyIqk+CPiKhMgj8iojIJ/oiIyiT4IyIqk+CPiKhMgj8iojIJ/oiIyiT4IyIqk+CPiKhMgj8iojIJ/oiIyiT4IyIqk+CPiKhMgj8iojIJ/oiIyiT4IyIqk+CPiKhMgj8iojJ9CX5Je0v6vaQ/SDquHzVERNSq58EvaQLwZWAfYCfgDZJ26nUdERG16scR/4uAP9i+xfZjwDeAmX2oIyKiSrLd2x1KrwP2tv22Mv0mYFfbRw9a70jgyDK5I/D7nhbaW5sC9/W7iFgpee/Gt9X9/dvG9pTBM9fsQyEaYt7TPn1snwWc1X45/Sdpru0Z/a4jVlzeu/Gt1vevH009dwJbd0xvBdzdhzoiIqrUj+D/JbCDpGdJegZwCHB5H+qIiKhSz5t6bD8h6Wjge8AE4Fzbv+l1HWNMFU1aq6m8d+Nble9fz0/uRkREf+XK3XFG0lAnxyMiupbgH3826HcBETG+JfjHEUn7A9+VtEGO/McWSc/sdw2x8iRt2O8aeinBP05I2gs4GXiv7cXOyZkxo4T+KZLWkZS/qXFG0quBCyVV8206/0nHgRL6FwC/An7b53Li6dYDXghsYfvJfhcT3St/W58CTrS9uN/19EqCf4yTtA/wz8CngT8B75G0fX+rCgBJkyWtafu/gf8Eppf5E/taWHSlhP7XgFuBX5d5VWRiFb/keCXp+cB+wNG2Pwf8GNgEOEjSdv2srXaSng18EZgj6ShgT+DlALYfL+uo8zHGDkl7A58BPgH8lKapbhvbT9bwfqUf/xhVjvQ/DbzJ9k0d818OvJpmYKmLbN/SlwIrJmkKzb//ZsBrgCeAQ4HdgXOAu4ArgYdt/6pfdcbQygHV+4Azbf9M0m40f1PrAKfZvl2SVufzaAn+MUjSnsDngXfY/nHpcbDM9kNl+cuBfYFHaa58/lO/aq2NpKnAKcAPbZ/XMX974Ks0w49MBF4JPAYcbntBH0qNIZQDqhOBt3WOGCBpV5oP8XWAz9m+Y3UO/zT1jDGS1qc5GrmyhP5U4LvA3w2sY/tqYA7NSKfVnJDqt9Lr4x7gv4DdJL25Y/GfgfnAxbZPAQ4EXpfQHztKm/7ngGNt/0bShuXbG7avAS4DHgJOkLTV6hr6kOAfUyRNA5bRHFGuX25L+S3g67avL+sIwPYc4GTb9/er3pqU0Pgh8BLb59K0C/+vcj8JyvuwCPjHMr3U9iN9KjcGKQdURwA/s/1zSZsCl9AxUrDta2nGELsDeLwvhfZIP8bjjyGUI48PAbfRNPM8CXwY+B1welnnb7562v5LH0qt1d8BzwE+Lumzts8vn8EvkzShNPv8CZjWxxpjCKUjxF00f1czJZ1Mc+vXczoPqNz4qaTrVvcP7QR/n3X8h1so6cfALsBRNGH/aeCdwBGS/tN27lvQPxcC29IcDR4laWJH+L9U0qM0J3Yn97HGGETSxsB7aJpwPkFzQHUMTRfOc8o6a3Ref7G6hz6kqWcsWBf++gFwETAPeAtN4F8HnA38A/A6SZv3q8gaSdpZ0s5l8gGak7U7AWcA75C0r+3zae4x8VLgEdvz+lNtdOrokrmI5hyZaL5Bz6Vp578deLekaTVedJfg7yNJ2wA/lbSLbZcTuW8GbqRpezwK+AUwG3guTS+e6AFJmwA3AFeU+0T/PfBRmvdAwNdpjvxn2j4H+Ijtpf2qN55mQnmU7e/Q/E0dAHyQ5gr4b9Lc/W+WpC37UmEfJfj7yPZtwEXAFyTtAZxP0zf/rTQXa21L08PnR8Axthf1q9balJO1e9CEw87A3jRXeT4MTLH9DeBS4FBJ69V0uf9YV07c/kHSZuWCrC1pmnuuoRnd9jiab2n/AawPrPZNO4OlH/8YIOlYmp48p9n+UMf8g4AZwCm2/9yv+momaXfgXJpzL6+juVDrTuCtwFo0f0MJ/TGmDLz2GZpbu34euMT2lzsugHycpj//hBo7SST4+6T0BFnWMX0scDDNBT+/7Zi//sCFW9EfkvalGS/pxbYfkvSsXDQ39pVhGb4NHF+urUDSBJpvci8DTrV9Xx9L7JsEfw9JegHwatufLNNDhf8hwNtt39inMmMIJfz/laYf/wNl3mp7ZefqolwF/0VgV9sPdsxf1/bD/ausv9LG3wMqaLrPvkzS8QC2l5UjEMr0qTSX/H9e0jNqGCxqvLD9bZq24SslrZHQHx/KhY7vA66VNLljfrWhDzni76ky5OsLaPoTXzvMkf/kgaPKGFvS9DY+SZoJfJzmnJlr/9BO8LesnEx6D3AasND2zZJmAB8Afm/742W9vwn/iBhd+dB+SoK/RaWp5ss0/fEvpblRx6U0A339Bjge+J7tL/WrxoioT4K/ZZLWAj4LPAO4mGa8l12ALcu8FwPvKhcBRUS0LsHfgtKUM5nmXqzfp7lA5Gyay/5Ps32npP8JbA/Mohl3/w/9qjci6pLgH2WS9qMZC+SbNP2Fb6a5YvBM4Kyy2gnAvWWYhokut+qLiOiFBP8okrQTzc0cZpVbum0M7EZzleeVNHdoOodmHJHjy5F/ugVGRE+lH/8o6OhvvzlwfQn9CWWYhR/SjPGyW+m1806app8noOlX1o+aI6JeCf7RMak8LgDWkbQO8GQ5mn+EpqnnRZL2sv0X20favqdv1UZE1RL8q6jc3eeEckL3DpreOgeVI/mB2yQuBH4APLjcDUVE9EiCf9VtCJhm5Mb1gHfTDLM8C9gIQNJhNCd65/epxoiIv8rJ3ZUkaaOB8fElPQd4A7AOzfDK29H03V9afnai+Rbw6/5UGxHxlAT/Sig3TTkd+A5wAc347Ka5Qnctmlvz3Q9sAWwC3GE7R/sRMSakqWfl3AdMo+mmuSvNHbL2p7mbzyKa5p5ptv9o+9qEfkSMJQn+lWD7BpphFx4FFgOvAp6kuS/r64H3A0dKmtivGiMilidNPatA0gtpLsw6xvZ5ZWz959F8EFxm++a+FhgRMYQE/yoq4f994KO2T+93PRERI1mz3wWMd7Z/WU72/lLSI7bP7XdNERHDyRH/KCn3033Y9u/7XUtExHAS/BERlUmvnoiIyiT4IyIqk+CPiKhMgj8iojIJ/oiIyiT4IyIqk+CPiKjM/wexWyofkzpSLgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Pour des raisons de visualisation , on va décoder les valeurs des classes : \n",
    "df_train_dec=df_train.replace({'type':{0:'HFT',1:'NON HFT',2:'MIX'}})\n",
    "# visualisation de la distribution du target\n",
    "plt.title('Distribution du Target')\n",
    "plt.ylabel('% des traders')\n",
    "(df_train_dec['type'].value_counts()/df_train_dec.shape[0]*100).plot.bar(rot=45,color='r',alpha=0.5)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7b59c23e",
   "metadata": {},
   "source": [
    "### Statistiques descriptives :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 587,
   "id": "6405916a",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>OTR</th>\n",
       "      <th>OCR</th>\n",
       "      <th>OMR</th>\n",
       "      <th>mean_time_two_events</th>\n",
       "      <th>90_p_time_two_events</th>\n",
       "      <th>75_p_lifetime_cancel</th>\n",
       "      <th>90_p_lifetime_cancel</th>\n",
       "      <th>NbTradeVenueMic</th>\n",
       "      <th>MeanNbTradesBySecond</th>\n",
       "      <th>min_dt_TV1</th>\n",
       "      <th>mean_dt_TV1</th>\n",
       "      <th>NbSecondWithAtLeatOneTrade</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>type</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>HFT</th>\n",
       "      <td>73.150343</td>\n",
       "      <td>31.947955</td>\n",
       "      <td>122.169833</td>\n",
       "      <td>872.861989</td>\n",
       "      <td>3252.076416</td>\n",
       "      <td>3742.003895</td>\n",
       "      <td>4802.965980</td>\n",
       "      <td>4.642871</td>\n",
       "      <td>2.442493</td>\n",
       "      <td>1140.896574</td>\n",
       "      <td>1878.587925</td>\n",
       "      <td>334.067432</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>MIX</th>\n",
       "      <td>16.248462</td>\n",
       "      <td>19.095098</td>\n",
       "      <td>181.528134</td>\n",
       "      <td>1270.441537</td>\n",
       "      <td>4956.524208</td>\n",
       "      <td>4038.193585</td>\n",
       "      <td>5494.844285</td>\n",
       "      <td>5.605962</td>\n",
       "      <td>2.579287</td>\n",
       "      <td>708.885694</td>\n",
       "      <td>1306.877468</td>\n",
       "      <td>266.380809</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NON HFT</th>\n",
       "      <td>15.247465</td>\n",
       "      <td>40.758430</td>\n",
       "      <td>102.910066</td>\n",
       "      <td>4739.855051</td>\n",
       "      <td>12622.162535</td>\n",
       "      <td>13420.314178</td>\n",
       "      <td>15694.911280</td>\n",
       "      <td>1.361570</td>\n",
       "      <td>2.958865</td>\n",
       "      <td>1119.273776</td>\n",
       "      <td>2544.944908</td>\n",
       "      <td>25.363415</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               OTR        OCR         OMR  mean_time_two_events  \\\n",
       "type                                                              \n",
       "HFT      73.150343  31.947955  122.169833            872.861989   \n",
       "MIX      16.248462  19.095098  181.528134           1270.441537   \n",
       "NON HFT  15.247465  40.758430  102.910066           4739.855051   \n",
       "\n",
       "         90_p_time_two_events  75_p_lifetime_cancel  90_p_lifetime_cancel  \\\n",
       "type                                                                        \n",
       "HFT               3252.076416           3742.003895           4802.965980   \n",
       "MIX               4956.524208           4038.193585           5494.844285   \n",
       "NON HFT          12622.162535          13420.314178          15694.911280   \n",
       "\n",
       "         NbTradeVenueMic  MeanNbTradesBySecond   min_dt_TV1  mean_dt_TV1  \\\n",
       "type                                                                       \n",
       "HFT             4.642871              2.442493  1140.896574  1878.587925   \n",
       "MIX             5.605962              2.579287   708.885694  1306.877468   \n",
       "NON HFT         1.361570              2.958865  1119.273776  2544.944908   \n",
       "\n",
       "         NbSecondWithAtLeatOneTrade  \n",
       "type                                 \n",
       "HFT                      334.067432  \n",
       "MIX                      266.380809  \n",
       "NON HFT                   25.363415  "
      ]
     },
     "execution_count": 587,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train_dec[['type','OTR', 'OCR', 'OMR', 'mean_time_two_events', '90_p_time_two_events',\n",
    "       '75_p_lifetime_cancel', '90_p_lifetime_cancel', 'NbTradeVenueMic',\n",
    "        'MeanNbTradesBySecond', 'min_dt_TV1',\n",
    "       'mean_dt_TV1', 'NbSecondWithAtLeatOneTrade']].groupby(['type']).mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a2f64056",
   "metadata": {},
   "source": [
    "### Visualization de la distribution de quelques variables selon la classe ( HFT , NON HFT ou MIX ):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 588,
   "id": "21b8d509",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plot_kde(x,y,titre):\n",
    "    plt.figure(figsize=(12,6))\n",
    "    sns.kdeplot(x[y=='HFT'],label='HFT')\n",
    "    sns.kdeplot(x[y=='MIX'],label='MIX')\n",
    "    sns.kdeplot(x[y=='NON HFT'],label='NON HFT')\n",
    "    plt.legend(fontsize=12)\n",
    "    plt.title(titre,fontsize=12)\n",
    "plot_kde(df_train_dec['mean_time_two_events'],df_train_dec['type'],'distribution du temps minimum entre 2 évènements')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 589,
   "id": "aa9225a1",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_kde(df_train_dec['NbTradeVenueMic'],df_train_dec['type'],'distribution du Nbre de trading venues')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 590,
   "id": "ddd79d7e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1800x648 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plt_histogram(x,y):\n",
    "    plt.hist(x[y=='HFT'],50,alpha=0.5,label='HFT')\n",
    "    plt.hist(x[y=='MIX'],50,alpha=0.5,label='MIX')\n",
    "    plt.hist(x[y=='NON HFT'],50,alpha=0.5,label='NON HFT')\n",
    "    plt.legend()\n",
    "    plt.show\n",
    "def plt_histogram_final(liste_variables):\n",
    "    plt.figure(figsize = (25,9))\n",
    "    for i in range(1,4):\n",
    "        plt.subplot(1,3,i)\n",
    "        plt.title('Distribution de'+' '+liste_variables[i-1],fontsize=20)\n",
    "        plt_histogram(df_train_dec[liste_variables[i-1]],df_train_dec['type'])\n",
    "    \n",
    "plt_histogram_final(['OTR','OCR','OMR'])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "26670519",
   "metadata": {},
   "source": [
    "### <u>6- Features selection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 591,
   "id": "efd005cc",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Comme 1ère étape , on va vérifier la corrélation des variables en utilisant la correlation map"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 592,
   "id": "2e87fa5b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "#T_5dc2b_row0_col0, #T_5dc2b_row1_col1, #T_5dc2b_row2_col2, #T_5dc2b_row3_col3, #T_5dc2b_row3_col5, #T_5dc2b_row4_col4, #T_5dc2b_row5_col3, #T_5dc2b_row5_col5, #T_5dc2b_row6_col6, #T_5dc2b_row7_col7, #T_5dc2b_row8_col8, #T_5dc2b_row9_col9, #T_5dc2b_row10_col10, #T_5dc2b_row11_col11, #T_5dc2b_row11_col13, #T_5dc2b_row12_col12, #T_5dc2b_row13_col11, #T_5dc2b_row13_col13, #T_5dc2b_row13_col15, #T_5dc2b_row14_col14, #T_5dc2b_row15_col13, #T_5dc2b_row15_col15, #T_5dc2b_row16_col16, #T_5dc2b_row17_col17, #T_5dc2b_row18_col18, #T_5dc2b_row19_col19, #T_5dc2b_row20_col20, #T_5dc2b_row21_col21, #T_5dc2b_row22_col22, #T_5dc2b_row23_col23, #T_5dc2b_row24_col24, #T_5dc2b_row25_col25, #T_5dc2b_row26_col26, #T_5dc2b_row26_col27, #T_5dc2b_row27_col26, #T_5dc2b_row27_col27, #T_5dc2b_row28_col28 {\n",
       "  background-color: #b40426;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row0_col1, #T_5dc2b_row6_col5, #T_5dc2b_row9_col11 {\n",
       "  background-color: #f7b396;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row0_col2 {\n",
       "  background-color: #536edd;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row0_col3, #T_5dc2b_row0_col5, #T_5dc2b_row0_col6, #T_5dc2b_row0_col7, #T_5dc2b_row2_col26, #T_5dc2b_row9_col1, #T_5dc2b_row19_col27, #T_5dc2b_row20_col4, #T_5dc2b_row21_col4, #T_5dc2b_row21_col17 {\n",
       "  background-color: #485fd1;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row0_col4, #T_5dc2b_row4_col0, #T_5dc2b_row20_col27, #T_5dc2b_row21_col3, #T_5dc2b_row21_col5, #T_5dc2b_row22_col2, #T_5dc2b_row25_col22 {\n",
       "  background-color: #4257c9;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row0_col8, #T_5dc2b_row0_col9, #T_5dc2b_row2_col1, #T_5dc2b_row9_col0, #T_5dc2b_row18_col21, #T_5dc2b_row19_col1, #T_5dc2b_row19_col3, #T_5dc2b_row19_col5, #T_5dc2b_row19_col7, #T_5dc2b_row19_col11, #T_5dc2b_row19_col12, #T_5dc2b_row19_col13, #T_5dc2b_row19_col14, #T_5dc2b_row19_col15, #T_5dc2b_row19_col16, #T_5dc2b_row19_col17, #T_5dc2b_row19_col23, #T_5dc2b_row20_col22, #T_5dc2b_row20_col24, #T_5dc2b_row21_col6, #T_5dc2b_row21_col8, #T_5dc2b_row21_col9, #T_5dc2b_row21_col10, #T_5dc2b_row21_col18, #T_5dc2b_row23_col2, #T_5dc2b_row23_col19, #T_5dc2b_row23_col20, #T_5dc2b_row23_col25, #T_5dc2b_row25_col1, #T_5dc2b_row25_col4, #T_5dc2b_row25_col6, #T_5dc2b_row25_col12, #T_5dc2b_row25_col14, #T_5dc2b_row25_col26, #T_5dc2b_row25_col27, #T_5dc2b_row25_col28 {\n",
       "  background-color: #3b4cc0;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row0_col10, #T_5dc2b_row10_col26, #T_5dc2b_row12_col25, #T_5dc2b_row17_col21, #T_5dc2b_row19_col28, #T_5dc2b_row24_col19, #T_5dc2b_row24_col25 {\n",
       "  background-color: #4f69d9;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row0_col11, #T_5dc2b_row0_col13, #T_5dc2b_row2_col9, #T_5dc2b_row3_col28, #T_5dc2b_row22_col6 {\n",
       "  background-color: #8caffe;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row0_col12, #T_5dc2b_row21_col28, #T_5dc2b_row23_col1, #T_5dc2b_row24_col6 {\n",
       "  background-color: #93b5fe;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row0_col14, #T_5dc2b_row0_col17, #T_5dc2b_row3_col20, #T_5dc2b_row5_col25 {\n",
       "  background-color: #8fb1fe;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row0_col15, #T_5dc2b_row3_col19, #T_5dc2b_row7_col25, #T_5dc2b_row7_col27, #T_5dc2b_row16_col0, #T_5dc2b_row17_col0, #T_5dc2b_row18_col0, #T_5dc2b_row18_col1 {\n",
       "  background-color: #89acfd;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row0_col16, #T_5dc2b_row0_col20, #T_5dc2b_row17_col26, #T_5dc2b_row18_col3 {\n",
       "  background-color: #97b8ff;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row0_col18, #T_5dc2b_row7_col28, #T_5dc2b_row19_col0, #T_5dc2b_row25_col18 {\n",
       "  background-color: #96b7ff;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row0_col19, #T_5dc2b_row0_col22, #T_5dc2b_row20_col18, #T_5dc2b_row26_col14, #T_5dc2b_row27_col16, #T_5dc2b_row28_col16 {\n",
       "  background-color: #b7cff9;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row0_col21, #T_5dc2b_row4_col20, #T_5dc2b_row6_col2, #T_5dc2b_row8_col26, #T_5dc2b_row27_col10, #T_5dc2b_row28_col10 {\n",
       "  background-color: #6c8ff1;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row0_col23, #T_5dc2b_row16_col3, #T_5dc2b_row26_col11, #T_5dc2b_row27_col11, #T_5dc2b_row27_col13 {\n",
       "  background-color: #bed2f6;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row0_col24, #T_5dc2b_row3_col23, #T_5dc2b_row18_col23, #T_5dc2b_row26_col16 {\n",
       "  background-color: #b5cdfa;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row0_col25, #T_5dc2b_row1_col23, #T_5dc2b_row7_col18, #T_5dc2b_row14_col3, #T_5dc2b_row18_col20, #T_5dc2b_row24_col8 {\n",
       "  background-color: #c7d7f0;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row0_col26, #T_5dc2b_row21_col22, #T_5dc2b_row24_col3 {\n",
       "  background-color: #779af7;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row0_col27, #T_5dc2b_row1_col27, #T_5dc2b_row2_col15, #T_5dc2b_row4_col27, #T_5dc2b_row22_col3, #T_5dc2b_row24_col5, #T_5dc2b_row28_col9 {\n",
       "  background-color: #7a9df8;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row0_col28, #T_5dc2b_row5_col27, #T_5dc2b_row8_col20, #T_5dc2b_row8_col28, #T_5dc2b_row12_col0, #T_5dc2b_row21_col26 {\n",
       "  background-color: #82a6fb;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row1_col0, #T_5dc2b_row4_col11, #T_5dc2b_row8_col17, #T_5dc2b_row9_col6 {\n",
       "  background-color: #f7a889;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row1_col2, #T_5dc2b_row9_col21, #T_5dc2b_row19_col9, #T_5dc2b_row20_col1, #T_5dc2b_row28_col2 {\n",
       "  background-color: #465ecf;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row1_col3, #T_5dc2b_row1_col5, #T_5dc2b_row4_col1 {\n",
       "  background-color: #4c66d6;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row1_col4, #T_5dc2b_row2_col3, #T_5dc2b_row2_col5, #T_5dc2b_row19_col21, #T_5dc2b_row24_col21, #T_5dc2b_row26_col10 {\n",
       "  background-color: #688aef;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row1_col6, #T_5dc2b_row3_col1, #T_5dc2b_row5_col1, #T_5dc2b_row6_col1, #T_5dc2b_row7_col1, #T_5dc2b_row11_col19, #T_5dc2b_row13_col19, #T_5dc2b_row15_col19, #T_5dc2b_row16_col2, #T_5dc2b_row22_col25 {\n",
       "  background-color: #5673e0;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row1_col7, #T_5dc2b_row2_col27, #T_5dc2b_row10_col1, #T_5dc2b_row22_col20 {\n",
       "  background-color: #4e68d8;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row1_col8, #T_5dc2b_row10_col27, #T_5dc2b_row15_col25 {\n",
       "  background-color: #5875e1;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row1_col9, #T_5dc2b_row6_col21, #T_5dc2b_row9_col26, #T_5dc2b_row12_col21, #T_5dc2b_row26_col3, #T_5dc2b_row28_col5 {\n",
       "  background-color: #6485ec;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row1_col10, #T_5dc2b_row2_col16, #T_5dc2b_row3_col21, #T_5dc2b_row8_col2, #T_5dc2b_row8_col19, #T_5dc2b_row21_col25, #T_5dc2b_row22_col10, #T_5dc2b_row23_col21, #T_5dc2b_row27_col9 {\n",
       "  background-color: #7699f6;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row1_col11, #T_5dc2b_row1_col13, #T_5dc2b_row1_col15, #T_5dc2b_row6_col22, #T_5dc2b_row14_col28 {\n",
       "  background-color: #b2ccfb;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row1_col12 {\n",
       "  background-color: #c5d6f2;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row1_col14, #T_5dc2b_row1_col17, #T_5dc2b_row3_col10, #T_5dc2b_row5_col18, #T_5dc2b_row16_col5 {\n",
       "  background-color: #c0d4f5;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row1_col16, #T_5dc2b_row11_col6, #T_5dc2b_row11_col7, #T_5dc2b_row13_col6, #T_5dc2b_row13_col7, #T_5dc2b_row14_col7, #T_5dc2b_row15_col7, #T_5dc2b_row23_col8 {\n",
       "  background-color: #cad8ef;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row1_col18, #T_5dc2b_row1_col24, #T_5dc2b_row5_col22, #T_5dc2b_row7_col22, #T_5dc2b_row15_col27, #T_5dc2b_row16_col1, #T_5dc2b_row25_col0 {\n",
       "  background-color: #adc9fd;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row1_col19, #T_5dc2b_row5_col26, #T_5dc2b_row11_col0, #T_5dc2b_row13_col0, #T_5dc2b_row18_col2, #T_5dc2b_row18_col27, #T_5dc2b_row19_col18, #T_5dc2b_row21_col1, #T_5dc2b_row22_col7 {\n",
       "  background-color: #80a3fa;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row1_col20, #T_5dc2b_row1_col28, #T_5dc2b_row7_col19, #T_5dc2b_row7_col26, #T_5dc2b_row23_col10, #T_5dc2b_row24_col10, #T_5dc2b_row28_col4 {\n",
       "  background-color: #86a9fc;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row1_col21, #T_5dc2b_row10_col2, #T_5dc2b_row12_col26, #T_5dc2b_row18_col7 {\n",
       "  background-color: #a5c3fe;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row1_col22, #T_5dc2b_row16_col27, #T_5dc2b_row24_col18 {\n",
       "  background-color: #a6c4fe;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row1_col25, #T_5dc2b_row5_col2, #T_5dc2b_row6_col20, #T_5dc2b_row6_col25, #T_5dc2b_row15_col0, #T_5dc2b_row28_col0 {\n",
       "  background-color: #7da0f9;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row1_col26, #T_5dc2b_row2_col23, #T_5dc2b_row4_col2, #T_5dc2b_row5_col21, #T_5dc2b_row7_col2, #T_5dc2b_row8_col25, #T_5dc2b_row26_col6, #T_5dc2b_row27_col6 {\n",
       "  background-color: #7597f6;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row2_col0, #T_5dc2b_row2_col22, #T_5dc2b_row6_col0, #T_5dc2b_row27_col3 {\n",
       "  background-color: #6384eb;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row2_col4, #T_5dc2b_row6_col19, #T_5dc2b_row21_col27 {\n",
       "  background-color: #85a8fc;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row2_col6, #T_5dc2b_row4_col25, #T_5dc2b_row11_col20, #T_5dc2b_row13_col20, #T_5dc2b_row20_col9 {\n",
       "  background-color: #6180e9;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row2_col7, #T_5dc2b_row15_col20, #T_5dc2b_row28_col3 {\n",
       "  background-color: #6282ea;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row2_col8, #T_5dc2b_row2_col11, #T_5dc2b_row2_col13, #T_5dc2b_row18_col26, #T_5dc2b_row27_col19, #T_5dc2b_row28_col6 {\n",
       "  background-color: #799cf8;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row2_col10, #T_5dc2b_row27_col14, #T_5dc2b_row28_col14 {\n",
       "  background-color: #bad0f8;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row2_col12, #T_5dc2b_row2_col14, #T_5dc2b_row9_col2, #T_5dc2b_row10_col22, #T_5dc2b_row15_col21, #T_5dc2b_row21_col15, #T_5dc2b_row22_col5, #T_5dc2b_row23_col7, #T_5dc2b_row26_col19 {\n",
       "  background-color: #7b9ff9;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row2_col17, #T_5dc2b_row4_col26, #T_5dc2b_row23_col5, #T_5dc2b_row26_col9, #T_5dc2b_row26_col20 {\n",
       "  background-color: #7295f4;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row2_col18, #T_5dc2b_row10_col3, #T_5dc2b_row18_col5, #T_5dc2b_row23_col6, #T_5dc2b_row26_col21, #T_5dc2b_row28_col21 {\n",
       "  background-color: #9abbff;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row2_col19, #T_5dc2b_row13_col18, #T_5dc2b_row18_col13, #T_5dc2b_row22_col26 {\n",
       "  background-color: #f3c8b2;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row2_col20 {\n",
       "  background-color: #c4d5f3;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row2_col21, #T_5dc2b_row5_col20, #T_5dc2b_row5_col28 {\n",
       "  background-color: #8db0fe;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row2_col24, #T_5dc2b_row16_col20, #T_5dc2b_row26_col5, #T_5dc2b_row27_col1, #T_5dc2b_row28_col1 {\n",
       "  background-color: #6788ee;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row2_col25, #T_5dc2b_row12_col3, #T_5dc2b_row16_col7, #T_5dc2b_row24_col9 {\n",
       "  background-color: #c6d6f1;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row2_col28, #T_5dc2b_row12_col20, #T_5dc2b_row15_col2, #T_5dc2b_row17_col19, #T_5dc2b_row25_col21 {\n",
       "  background-color: #5d7ce6;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row3_col0, #T_5dc2b_row5_col0, #T_5dc2b_row9_col27, #T_5dc2b_row14_col21, #T_5dc2b_row26_col25, #T_5dc2b_row27_col7, #T_5dc2b_row27_col8 {\n",
       "  background-color: #6e90f2;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row3_col2, #T_5dc2b_row3_col26, #T_5dc2b_row9_col28, #T_5dc2b_row22_col1, #T_5dc2b_row24_col7, #T_5dc2b_row26_col4 {\n",
       "  background-color: #7ea1fa;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row3_col4 {\n",
       "  background-color: #f7b89c;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row3_col6, #T_5dc2b_row4_col18, #T_5dc2b_row8_col18, #T_5dc2b_row10_col12, #T_5dc2b_row15_col18, #T_5dc2b_row18_col15 {\n",
       "  background-color: #f7bca1;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row3_col7, #T_5dc2b_row7_col3 {\n",
       "  background-color: #d0473d;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row3_col8, #T_5dc2b_row15_col24, #T_5dc2b_row24_col11, #T_5dc2b_row24_col28, #T_5dc2b_row28_col23 {\n",
       "  background-color: #f2c9b4;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row3_col9, #T_5dc2b_row4_col23, #T_5dc2b_row5_col11, #T_5dc2b_row22_col14 {\n",
       "  background-color: #e6d7cf;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row3_col11, #T_5dc2b_row3_col13, #T_5dc2b_row5_col13, #T_5dc2b_row5_col15, #T_5dc2b_row6_col10, #T_5dc2b_row6_col17, #T_5dc2b_row7_col12, #T_5dc2b_row10_col15 {\n",
       "  background-color: #e5d8d1;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row3_col12, #T_5dc2b_row16_col24, #T_5dc2b_row19_col2, #T_5dc2b_row23_col27 {\n",
       "  background-color: #e3d9d3;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row3_col14, #T_5dc2b_row5_col14, #T_5dc2b_row7_col11, #T_5dc2b_row7_col13, #T_5dc2b_row7_col14, #T_5dc2b_row7_col15 {\n",
       "  background-color: #e2dad5;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row3_col15, #T_5dc2b_row5_col12, #T_5dc2b_row8_col23, #T_5dc2b_row12_col22, #T_5dc2b_row15_col10, #T_5dc2b_row22_col12 {\n",
       "  background-color: #e4d9d2;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row3_col16, #T_5dc2b_row10_col13 {\n",
       "  background-color: #dbdcde;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row3_col17, #T_5dc2b_row10_col20, #T_5dc2b_row11_col5, #T_5dc2b_row13_col5, #T_5dc2b_row23_col17, #T_5dc2b_row24_col4, #T_5dc2b_row24_col17 {\n",
       "  background-color: #cedaeb;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row3_col18, #T_5dc2b_row9_col22, #T_5dc2b_row17_col7, #T_5dc2b_row20_col10, #T_5dc2b_row27_col12, #T_5dc2b_row28_col12 {\n",
       "  background-color: #bfd3f6;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row3_col22, #T_5dc2b_row10_col25, #T_5dc2b_row12_col27, #T_5dc2b_row18_col25, #T_5dc2b_row27_col17, #T_5dc2b_row28_col17 {\n",
       "  background-color: #abc8fd;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row3_col24, #T_5dc2b_row13_col27, #T_5dc2b_row18_col24 {\n",
       "  background-color: #aec9fc;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row3_col25, #T_5dc2b_row11_col1, #T_5dc2b_row13_col1, #T_5dc2b_row21_col23, #T_5dc2b_row28_col18 {\n",
       "  background-color: #90b2fe;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row3_col27, #T_5dc2b_row11_col21, #T_5dc2b_row13_col21, #T_5dc2b_row14_col0, #T_5dc2b_row21_col11, #T_5dc2b_row21_col13, #T_5dc2b_row24_col1, #T_5dc2b_row27_col0, #T_5dc2b_row27_col4 {\n",
       "  background-color: #81a4fb;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row4_col3, #T_5dc2b_row10_col14, #T_5dc2b_row11_col24, #T_5dc2b_row13_col24, #T_5dc2b_row18_col8 {\n",
       "  background-color: #f4c6af;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row4_col5, #T_5dc2b_row12_col10, #T_5dc2b_row28_col22 {\n",
       "  background-color: #f6bea4;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row4_col6, #T_5dc2b_row8_col6, #T_5dc2b_row13_col16, #T_5dc2b_row16_col13 {\n",
       "  background-color: #e36b54;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row4_col7, #T_5dc2b_row8_col14, #T_5dc2b_row11_col17, #T_5dc2b_row14_col18, #T_5dc2b_row18_col9, #T_5dc2b_row18_col14, #T_5dc2b_row25_col19 {\n",
       "  background-color: #f4987a;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row4_col8, #T_5dc2b_row14_col15, #T_5dc2b_row15_col14 {\n",
       "  background-color: #bb1b2c;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row4_col9, #T_5dc2b_row9_col4, #T_5dc2b_row13_col14, #T_5dc2b_row14_col13, #T_5dc2b_row14_col16, #T_5dc2b_row16_col14 {\n",
       "  background-color: #c32e31;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row4_col10 {\n",
       "  background-color: #f5a081;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row4_col12 {\n",
       "  background-color: #ef886b;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row4_col13, #T_5dc2b_row16_col8 {\n",
       "  background-color: #f7a688;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row4_col14, #T_5dc2b_row4_col16, #T_5dc2b_row7_col4 {\n",
       "  background-color: #f18d6f;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row4_col15, #T_5dc2b_row6_col9, #T_5dc2b_row7_col8, #T_5dc2b_row10_col17 {\n",
       "  background-color: #f59d7e;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row4_col17, #T_5dc2b_row8_col12, #T_5dc2b_row17_col11 {\n",
       "  background-color: #f39778;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row4_col19, #T_5dc2b_row21_col12, #T_5dc2b_row27_col25 {\n",
       "  background-color: #6a8bef;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row4_col21, #T_5dc2b_row8_col21, #T_5dc2b_row16_col19, #T_5dc2b_row16_col25 {\n",
       "  background-color: #5572df;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row4_col22, #T_5dc2b_row6_col23, #T_5dc2b_row12_col7, #T_5dc2b_row15_col3, #T_5dc2b_row23_col9 {\n",
       "  background-color: #ccd9ed;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row4_col24, #T_5dc2b_row8_col24, #T_5dc2b_row9_col5, #T_5dc2b_row13_col10 {\n",
       "  background-color: #dadce0;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row4_col28 {\n",
       "  background-color: #8badfd;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row5_col4, #T_5dc2b_row5_col6, #T_5dc2b_row9_col13 {\n",
       "  background-color: #f7b093;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row5_col7, #T_5dc2b_row7_col5, #T_5dc2b_row12_col17 {\n",
       "  background-color: #c83836;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row5_col8, #T_5dc2b_row27_col24, #T_5dc2b_row28_col24 {\n",
       "  background-color: #f5c2aa;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row5_col9 {\n",
       "  background-color: #ead5c9;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row5_col10 {\n",
       "  background-color: #c3d5f4;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row5_col16 {\n",
       "  background-color: #dcdddd;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row5_col17 {\n",
       "  background-color: #d1dae9;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row5_col19, #T_5dc2b_row6_col27, #T_5dc2b_row7_col20, #T_5dc2b_row18_col28 {\n",
       "  background-color: #88abfd;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row5_col23, #T_5dc2b_row10_col19, #T_5dc2b_row11_col28, #T_5dc2b_row12_col28, #T_5dc2b_row13_col28, #T_5dc2b_row15_col28 {\n",
       "  background-color: #b6cefa;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row5_col24, #T_5dc2b_row11_col27, #T_5dc2b_row21_col20 {\n",
       "  background-color: #afcafc;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row6_col3 {\n",
       "  background-color: #f6bfa6;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row6_col4 {\n",
       "  background-color: #e0654f;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row6_col7 {\n",
       "  background-color: #df634e;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row6_col8, #T_5dc2b_row10_col18, #T_5dc2b_row18_col10, #T_5dc2b_row18_col16 {\n",
       "  background-color: #e16751;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row6_col11, #T_5dc2b_row6_col13 {\n",
       "  background-color: #dedcdb;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row6_col12, #T_5dc2b_row23_col12, #T_5dc2b_row23_col14, #T_5dc2b_row24_col12 {\n",
       "  background-color: #ebd3c6;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row6_col14, #T_5dc2b_row14_col22 {\n",
       "  background-color: #e7d7ce;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row6_col15 {\n",
       "  background-color: #e1dad6;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row6_col16, #T_5dc2b_row16_col23, #T_5dc2b_row23_col28 {\n",
       "  background-color: #ead4c8;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row6_col18, #T_5dc2b_row9_col3 {\n",
       "  background-color: #d5dbe5;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row6_col24, #T_5dc2b_row26_col15, #T_5dc2b_row28_col13, #T_5dc2b_row28_col15 {\n",
       "  background-color: #bbd1f8;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row6_col26, #T_5dc2b_row9_col20, #T_5dc2b_row21_col19, #T_5dc2b_row26_col0 {\n",
       "  background-color: #84a7fc;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row6_col28, #T_5dc2b_row18_col19, #T_5dc2b_row27_col21 {\n",
       "  background-color: #98b9ff;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row7_col0, #T_5dc2b_row26_col8 {\n",
       "  background-color: #6b8df0;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row7_col6 {\n",
       "  background-color: #de614d;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row7_col9, #T_5dc2b_row26_col24 {\n",
       "  background-color: #f5c4ac;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row7_col10, #T_5dc2b_row8_col22 {\n",
       "  background-color: #cfdaea;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row7_col16, #T_5dc2b_row9_col23, #T_5dc2b_row23_col16, #T_5dc2b_row23_col26 {\n",
       "  background-color: #e0dbd8;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row7_col17, #T_5dc2b_row16_col22 {\n",
       "  background-color: #d7dce3;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row7_col21, #T_5dc2b_row9_col19, #T_5dc2b_row9_col25, #T_5dc2b_row21_col14, #T_5dc2b_row23_col3, #T_5dc2b_row26_col7, #T_5dc2b_row27_col20, #T_5dc2b_row28_col7 {\n",
       "  background-color: #6f92f3;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row7_col23, #T_5dc2b_row18_col6, #T_5dc2b_row22_col8, #T_5dc2b_row26_col12, #T_5dc2b_row26_col13, #T_5dc2b_row27_col15, #T_5dc2b_row28_col11 {\n",
       "  background-color: #bcd2f7;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row7_col24, #T_5dc2b_row10_col7, #T_5dc2b_row16_col28, #T_5dc2b_row17_col3 {\n",
       "  background-color: #b1cbfc;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row8_col0, #T_5dc2b_row8_col1, #T_5dc2b_row20_col11, #T_5dc2b_row20_col13, #T_5dc2b_row20_col14, #T_5dc2b_row20_col15, #T_5dc2b_row24_col20, #T_5dc2b_row25_col9 {\n",
       "  background-color: #4961d2;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row8_col3 {\n",
       "  background-color: #edd1c2;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row8_col4, #T_5dc2b_row12_col16, #T_5dc2b_row16_col12 {\n",
       "  background-color: #ba162b;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row8_col5, #T_5dc2b_row12_col23, #T_5dc2b_row13_col22, #T_5dc2b_row23_col13, #T_5dc2b_row26_col23 {\n",
       "  background-color: #f1ccb8;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row8_col7, #T_5dc2b_row8_col10, #T_5dc2b_row10_col4 {\n",
       "  background-color: #f6a586;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row8_col9, #T_5dc2b_row9_col8 {\n",
       "  background-color: #ca3b37;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row8_col11 {\n",
       "  background-color: #f7ac8e;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row8_col13, #T_5dc2b_row13_col4 {\n",
       "  background-color: #f7aa8c;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row8_col15, #T_5dc2b_row9_col15 {\n",
       "  background-color: #f6a385;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row8_col16, #T_5dc2b_row17_col10 {\n",
       "  background-color: #f59c7d;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row8_col27, #T_5dc2b_row20_col0, #T_5dc2b_row21_col24, #T_5dc2b_row22_col21, #T_5dc2b_row28_col19 {\n",
       "  background-color: #7396f5;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row9_col7, #T_5dc2b_row24_col26 {\n",
       "  background-color: #efcfbf;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row9_col10, #T_5dc2b_row11_col16, #T_5dc2b_row16_col11 {\n",
       "  background-color: #e46e56;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row9_col12, #T_5dc2b_row16_col9, #T_5dc2b_row17_col15 {\n",
       "  background-color: #ec8165;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row9_col14 {\n",
       "  background-color: #f08a6c;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row9_col16 {\n",
       "  background-color: #ea7b60;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row9_col17 {\n",
       "  background-color: #eb7d62;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row9_col18, #T_5dc2b_row13_col17 {\n",
       "  background-color: #f39577;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row9_col24, #T_5dc2b_row10_col6 {\n",
       "  background-color: #d3dbe7;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row10_col0, #T_5dc2b_row25_col5 {\n",
       "  background-color: #4358cb;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row10_col5, #T_5dc2b_row16_col26, #T_5dc2b_row23_col0, #T_5dc2b_row23_col18, #T_5dc2b_row25_col2 {\n",
       "  background-color: #9ebeff;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row10_col8, #T_5dc2b_row16_col10, #T_5dc2b_row17_col8 {\n",
       "  background-color: #f7af91;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row10_col9 {\n",
       "  background-color: #e67259;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row10_col11 {\n",
       "  background-color: #d9dce1;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row10_col16, #T_5dc2b_row11_col4, #T_5dc2b_row15_col8 {\n",
       "  background-color: #f7ad90;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row10_col21, #T_5dc2b_row20_col6 {\n",
       "  background-color: #3c4ec2;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row10_col23 {\n",
       "  background-color: #a1c0ff;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row10_col24, #T_5dc2b_row15_col1, #T_5dc2b_row26_col18 {\n",
       "  background-color: #92b4fe;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row10_col28, #T_5dc2b_row26_col1, #T_5dc2b_row27_col5, #T_5dc2b_row28_col20 {\n",
       "  background-color: #6687ed;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row11_col2, #T_5dc2b_row13_col2, #T_5dc2b_row14_col2, #T_5dc2b_row17_col2, #T_5dc2b_row21_col16, #T_5dc2b_row28_col25 {\n",
       "  background-color: #5b7ae5;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row11_col3, #T_5dc2b_row13_col3, #T_5dc2b_row15_col5, #T_5dc2b_row15_col6 {\n",
       "  background-color: #cdd9ec;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row11_col8 {\n",
       "  background-color: #f7b599;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row11_col9 {\n",
       "  background-color: #f7b79b;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row11_col10, #T_5dc2b_row12_col6, #T_5dc2b_row16_col6 {\n",
       "  background-color: #d8dce2;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row11_col12, #T_5dc2b_row12_col11 {\n",
       "  background-color: #cf453c;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row11_col14, #T_5dc2b_row14_col11, #T_5dc2b_row15_col12 {\n",
       "  background-color: #c43032;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row11_col15, #T_5dc2b_row12_col14, #T_5dc2b_row14_col12, #T_5dc2b_row15_col11, #T_5dc2b_row27_col28, #T_5dc2b_row28_col27 {\n",
       "  background-color: #b50927;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row11_col18, #T_5dc2b_row18_col11, #T_5dc2b_row20_col19, #T_5dc2b_row24_col13 {\n",
       "  background-color: #f2cab5;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row11_col22, #T_5dc2b_row14_col23, #T_5dc2b_row19_col20, #T_5dc2b_row23_col11, #T_5dc2b_row27_col23 {\n",
       "  background-color: #f2cbb7;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row11_col23, #T_5dc2b_row13_col23, #T_5dc2b_row22_col28 {\n",
       "  background-color: #f5c1a9;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row11_col25, #T_5dc2b_row12_col2, #T_5dc2b_row13_col25, #T_5dc2b_row22_col19 {\n",
       "  background-color: #5977e3;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row11_col26, #T_5dc2b_row13_col26, #T_5dc2b_row17_col28, #T_5dc2b_row22_col0 {\n",
       "  background-color: #aac7fd;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row12_col1 {\n",
       "  background-color: #a3c2fe;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row12_col4, #T_5dc2b_row14_col9 {\n",
       "  background-color: #f18f71;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row12_col5, #T_5dc2b_row14_col5, #T_5dc2b_row17_col22 {\n",
       "  background-color: #c9d7f0;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row12_col8, #T_5dc2b_row14_col8, #T_5dc2b_row15_col4 {\n",
       "  background-color: #f6a283;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row12_col9, #T_5dc2b_row12_col18 {\n",
       "  background-color: #ee8669;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row12_col13 {\n",
       "  background-color: #cd423b;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row12_col15, #T_5dc2b_row22_col23 {\n",
       "  background-color: #c53334;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row12_col19, #T_5dc2b_row14_col25, #T_5dc2b_row20_col17 {\n",
       "  background-color: #516ddb;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row12_col24 {\n",
       "  background-color: #edd2c3;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row13_col8, #T_5dc2b_row13_col9 {\n",
       "  background-color: #f7b497;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row13_col12 {\n",
       "  background-color: #cc403a;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row14_col1, #T_5dc2b_row17_col27, #T_5dc2b_row18_col22, #T_5dc2b_row19_col10 {\n",
       "  background-color: #9fbfff;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row14_col4, #T_5dc2b_row16_col4 {\n",
       "  background-color: #f29274;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row14_col6, #T_5dc2b_row23_col4 {\n",
       "  background-color: #d4dbe6;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row14_col10, #T_5dc2b_row22_col27 {\n",
       "  background-color: #f3c7b1;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row14_col17, #T_5dc2b_row17_col14, #T_5dc2b_row17_col18 {\n",
       "  background-color: #d55042;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row14_col19 {\n",
       "  background-color: #5470de;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row14_col20, #T_5dc2b_row17_col25, #T_5dc2b_row21_col0 {\n",
       "  background-color: #5f7fe8;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row14_col24, #T_5dc2b_row22_col15 {\n",
       "  background-color: #eed0c0;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row14_col26, #T_5dc2b_row24_col0 {\n",
       "  background-color: #a2c1ff;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row14_col27, #T_5dc2b_row17_col1, #T_5dc2b_row26_col17 {\n",
       "  background-color: #a7c5fe;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row15_col9 {\n",
       "  background-color: #f7a98b;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row15_col16 {\n",
       "  background-color: #d95847;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row15_col17 {\n",
       "  background-color: #ed8366;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row15_col22, #T_5dc2b_row23_col15 {\n",
       "  background-color: #efcebd;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row15_col23 {\n",
       "  background-color: #f4c5ad;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row15_col26 {\n",
       "  background-color: #a9c6fd;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row16_col15 {\n",
       "  background-color: #da5a49;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row16_col17, #T_5dc2b_row17_col16 {\n",
       "  background-color: #b8122a;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row16_col18 {\n",
       "  background-color: #e26952;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row16_col21 {\n",
       "  background-color: #5a78e4;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row17_col4, #T_5dc2b_row19_col25 {\n",
       "  background-color: #f49a7b;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row17_col5, #T_5dc2b_row22_col9 {\n",
       "  background-color: #b3cdfb;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row17_col6, #T_5dc2b_row17_col24, #T_5dc2b_row22_col16 {\n",
       "  background-color: #d6dce4;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row17_col9 {\n",
       "  background-color: #ec7f63;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row17_col12, #T_5dc2b_row23_col22 {\n",
       "  background-color: #c73635;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row17_col13 {\n",
       "  background-color: #f39475;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row17_col20, #T_5dc2b_row21_col2, #T_5dc2b_row28_col8 {\n",
       "  background-color: #7093f3;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row17_col23, #T_5dc2b_row24_col16 {\n",
       "  background-color: #dfdbd9;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row18_col4 {\n",
       "  background-color: #f5c0a7;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row18_col12 {\n",
       "  background-color: #ee8468;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row18_col17 {\n",
       "  background-color: #d65244;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row19_col4, #T_5dc2b_row19_col22, #T_5dc2b_row20_col26, #T_5dc2b_row24_col2 {\n",
       "  background-color: #4055c8;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row19_col6, #T_5dc2b_row19_col8, #T_5dc2b_row20_col7, #T_5dc2b_row25_col7, #T_5dc2b_row25_col8, #T_5dc2b_row25_col11, #T_5dc2b_row25_col13, #T_5dc2b_row25_col24 {\n",
       "  background-color: #3f53c6;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row19_col24, #T_5dc2b_row21_col7, #T_5dc2b_row25_col15, #T_5dc2b_row25_col16, #T_5dc2b_row25_col23 {\n",
       "  background-color: #3d50c3;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row19_col26, #T_5dc2b_row20_col3, #T_5dc2b_row20_col28, #T_5dc2b_row27_col2 {\n",
       "  background-color: #455cce;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row20_col2, #T_5dc2b_row20_col21, #T_5dc2b_row22_col18 {\n",
       "  background-color: #9bbcff;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row20_col5, #T_5dc2b_row25_col3, #T_5dc2b_row26_col2 {\n",
       "  background-color: #445acc;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row20_col8, #T_5dc2b_row20_col16 {\n",
       "  background-color: #506bda;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row20_col12 {\n",
       "  background-color: #4a63d3;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row20_col23, #T_5dc2b_row25_col17 {\n",
       "  background-color: #3e51c5;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row20_col25, #T_5dc2b_row25_col20 {\n",
       "  background-color: #f29072;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row22_col4, #T_5dc2b_row22_col17 {\n",
       "  background-color: #c1d4f4;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row22_col11, #T_5dc2b_row22_col13, #T_5dc2b_row24_col15 {\n",
       "  background-color: #f1cdba;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row22_col24, #T_5dc2b_row24_col22, #T_5dc2b_row26_col28, #T_5dc2b_row28_col26 {\n",
       "  background-color: #b70d28;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row23_col24, #T_5dc2b_row24_col23 {\n",
       "  background-color: #bd1f2d;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5dc2b_row24_col14 {\n",
       "  background-color: #ecd3c5;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row24_col27 {\n",
       "  background-color: #f0cdbb;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row25_col10, #T_5dc2b_row27_col18 {\n",
       "  background-color: #94b6ff;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5dc2b_row26_col22, #T_5dc2b_row27_col22 {\n",
       "  background-color: #f6bda2;\n",
       "  color: #000000;\n",
       "}\n",
       "</style>\n",
       "<table id=\"T_5dc2b_\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th class=\"col_heading level0 col0\" >OTR</th>\n",
       "      <th class=\"col_heading level0 col1\" >OCR</th>\n",
       "      <th class=\"col_heading level0 col2\" >OMR</th>\n",
       "      <th class=\"col_heading level0 col3\" >min_time_two_events</th>\n",
       "      <th class=\"col_heading level0 col4\" >mean_time_two_events</th>\n",
       "      <th class=\"col_heading level0 col5\" >10_p_time_two_events</th>\n",
       "      <th class=\"col_heading level0 col6\" >med_time_two_events</th>\n",
       "      <th class=\"col_heading level0 col7\" >25_p_time_two_events</th>\n",
       "      <th class=\"col_heading level0 col8\" >75_p_time_two_events</th>\n",
       "      <th class=\"col_heading level0 col9\" >90_p_time_two_events</th>\n",
       "      <th class=\"col_heading level0 col10\" >max_time_two_events</th>\n",
       "      <th class=\"col_heading level0 col11\" >min_lifetime_cancel</th>\n",
       "      <th class=\"col_heading level0 col12\" >mean_lifetime_cancel</th>\n",
       "      <th class=\"col_heading level0 col13\" >10_p_lifetime_cancel</th>\n",
       "      <th class=\"col_heading level0 col14\" >med_lifetime_cancel</th>\n",
       "      <th class=\"col_heading level0 col15\" >25_p_lifetime_cancel</th>\n",
       "      <th class=\"col_heading level0 col16\" >75_p_lifetime_cancel</th>\n",
       "      <th class=\"col_heading level0 col17\" >90_p_lifetime_cancel</th>\n",
       "      <th class=\"col_heading level0 col18\" >max_lifetime_cancel</th>\n",
       "      <th class=\"col_heading level0 col19\" >NbTradeVenueMic</th>\n",
       "      <th class=\"col_heading level0 col20\" >MaxNbTradesBySecond</th>\n",
       "      <th class=\"col_heading level0 col21\" >MeanNbTradesBySecond</th>\n",
       "      <th class=\"col_heading level0 col22\" >min_dt_TV1</th>\n",
       "      <th class=\"col_heading level0 col23\" >mean_dt_TV1</th>\n",
       "      <th class=\"col_heading level0 col24\" >med_dt_TV1</th>\n",
       "      <th class=\"col_heading level0 col25\" >NbSecondWithAtLeatOneTrade</th>\n",
       "      <th class=\"col_heading level0 col26\" >moy_min_dt_TVi_TVj</th>\n",
       "      <th class=\"col_heading level0 col27\" >moy_med_dt_TVi_TVj</th>\n",
       "      <th class=\"col_heading level0 col28\" >moy_mean_dt_TVi_TVj</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row0\" class=\"row_heading level0 row0\" >OTR</th>\n",
       "      <td id=\"T_5dc2b_row0_col0\" class=\"data row0 col0\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row0_col1\" class=\"data row0 col1\" >0.62</td>\n",
       "      <td id=\"T_5dc2b_row0_col2\" class=\"data row0 col2\" >-0.13</td>\n",
       "      <td id=\"T_5dc2b_row0_col3\" class=\"data row0 col3\" >-0.09</td>\n",
       "      <td id=\"T_5dc2b_row0_col4\" class=\"data row0 col4\" >-0.27</td>\n",
       "      <td id=\"T_5dc2b_row0_col5\" class=\"data row0 col5\" >-0.09</td>\n",
       "      <td id=\"T_5dc2b_row0_col6\" class=\"data row0 col6\" >-0.13</td>\n",
       "      <td id=\"T_5dc2b_row0_col7\" class=\"data row0 col7\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row0_col8\" class=\"data row0 col8\" >-0.24</td>\n",
       "      <td id=\"T_5dc2b_row0_col9\" class=\"data row0 col9\" >-0.30</td>\n",
       "      <td id=\"T_5dc2b_row0_col10\" class=\"data row0 col10\" >-0.26</td>\n",
       "      <td id=\"T_5dc2b_row0_col11\" class=\"data row0 col11\" >-0.02</td>\n",
       "      <td id=\"T_5dc2b_row0_col12\" class=\"data row0 col12\" >-0.01</td>\n",
       "      <td id=\"T_5dc2b_row0_col13\" class=\"data row0 col13\" >-0.02</td>\n",
       "      <td id=\"T_5dc2b_row0_col14\" class=\"data row0 col14\" >-0.02</td>\n",
       "      <td id=\"T_5dc2b_row0_col15\" class=\"data row0 col15\" >-0.03</td>\n",
       "      <td id=\"T_5dc2b_row0_col16\" class=\"data row0 col16\" >0.01</td>\n",
       "      <td id=\"T_5dc2b_row0_col17\" class=\"data row0 col17\" >0.01</td>\n",
       "      <td id=\"T_5dc2b_row0_col18\" class=\"data row0 col18\" >0.01</td>\n",
       "      <td id=\"T_5dc2b_row0_col19\" class=\"data row0 col19\" >0.06</td>\n",
       "      <td id=\"T_5dc2b_row0_col20\" class=\"data row0 col20\" >-0.07</td>\n",
       "      <td id=\"T_5dc2b_row0_col21\" class=\"data row0 col21\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row0_col22\" class=\"data row0 col22\" >0.14</td>\n",
       "      <td id=\"T_5dc2b_row0_col23\" class=\"data row0 col23\" >0.09</td>\n",
       "      <td id=\"T_5dc2b_row0_col24\" class=\"data row0 col24\" >0.10</td>\n",
       "      <td id=\"T_5dc2b_row0_col25\" class=\"data row0 col25\" >0.15</td>\n",
       "      <td id=\"T_5dc2b_row0_col26\" class=\"data row0 col26\" >-0.01</td>\n",
       "      <td id=\"T_5dc2b_row0_col27\" class=\"data row0 col27\" >-0.02</td>\n",
       "      <td id=\"T_5dc2b_row0_col28\" class=\"data row0 col28\" >-0.03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row1\" class=\"row_heading level0 row1\" >OCR</th>\n",
       "      <td id=\"T_5dc2b_row1_col0\" class=\"data row1 col0\" >0.62</td>\n",
       "      <td id=\"T_5dc2b_row1_col1\" class=\"data row1 col1\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row1_col2\" class=\"data row1 col2\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row1_col3\" class=\"data row1 col3\" >-0.07</td>\n",
       "      <td id=\"T_5dc2b_row1_col4\" class=\"data row1 col4\" >-0.11</td>\n",
       "      <td id=\"T_5dc2b_row1_col5\" class=\"data row1 col5\" >-0.07</td>\n",
       "      <td id=\"T_5dc2b_row1_col6\" class=\"data row1 col6\" >-0.07</td>\n",
       "      <td id=\"T_5dc2b_row1_col7\" class=\"data row1 col7\" >-0.07</td>\n",
       "      <td id=\"T_5dc2b_row1_col8\" class=\"data row1 col8\" >-0.12</td>\n",
       "      <td id=\"T_5dc2b_row1_col9\" class=\"data row1 col9\" >-0.13</td>\n",
       "      <td id=\"T_5dc2b_row1_col10\" class=\"data row1 col10\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row1_col11\" class=\"data row1 col11\" >0.13</td>\n",
       "      <td id=\"T_5dc2b_row1_col12\" class=\"data row1 col12\" >0.19</td>\n",
       "      <td id=\"T_5dc2b_row1_col13\" class=\"data row1 col13\" >0.13</td>\n",
       "      <td id=\"T_5dc2b_row1_col14\" class=\"data row1 col14\" >0.18</td>\n",
       "      <td id=\"T_5dc2b_row1_col15\" class=\"data row1 col15\" >0.13</td>\n",
       "      <td id=\"T_5dc2b_row1_col16\" class=\"data row1 col16\" >0.22</td>\n",
       "      <td id=\"T_5dc2b_row1_col17\" class=\"data row1 col17\" >0.20</td>\n",
       "      <td id=\"T_5dc2b_row1_col18\" class=\"data row1 col18\" >0.10</td>\n",
       "      <td id=\"T_5dc2b_row1_col19\" class=\"data row1 col19\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row1_col20\" class=\"data row1 col20\" >-0.13</td>\n",
       "      <td id=\"T_5dc2b_row1_col21\" class=\"data row1 col21\" >0.07</td>\n",
       "      <td id=\"T_5dc2b_row1_col22\" class=\"data row1 col22\" >0.06</td>\n",
       "      <td id=\"T_5dc2b_row1_col23\" class=\"data row1 col23\" >0.14</td>\n",
       "      <td id=\"T_5dc2b_row1_col24\" class=\"data row1 col24\" >0.07</td>\n",
       "      <td id=\"T_5dc2b_row1_col25\" class=\"data row1 col25\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row1_col26\" class=\"data row1 col26\" >-0.02</td>\n",
       "      <td id=\"T_5dc2b_row1_col27\" class=\"data row1 col27\" >-0.01</td>\n",
       "      <td id=\"T_5dc2b_row1_col28\" class=\"data row1 col28\" >-0.01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row2\" class=\"row_heading level0 row2\" >OMR</th>\n",
       "      <td id=\"T_5dc2b_row2_col0\" class=\"data row2 col0\" >-0.13</td>\n",
       "      <td id=\"T_5dc2b_row2_col1\" class=\"data row2 col1\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row2_col2\" class=\"data row2 col2\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row2_col3\" class=\"data row2 col3\" >0.03</td>\n",
       "      <td id=\"T_5dc2b_row2_col4\" class=\"data row2 col4\" >-0.01</td>\n",
       "      <td id=\"T_5dc2b_row2_col5\" class=\"data row2 col5\" >0.02</td>\n",
       "      <td id=\"T_5dc2b_row2_col6\" class=\"data row2 col6\" >-0.04</td>\n",
       "      <td id=\"T_5dc2b_row2_col7\" class=\"data row2 col7\" >-0.00</td>\n",
       "      <td id=\"T_5dc2b_row2_col8\" class=\"data row2 col8\" >-0.00</td>\n",
       "      <td id=\"T_5dc2b_row2_col9\" class=\"data row2 col9\" >0.02</td>\n",
       "      <td id=\"T_5dc2b_row2_col10\" class=\"data row2 col10\" >0.16</td>\n",
       "      <td id=\"T_5dc2b_row2_col11\" class=\"data row2 col11\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row2_col12\" class=\"data row2 col12\" >-0.11</td>\n",
       "      <td id=\"T_5dc2b_row2_col13\" class=\"data row2 col13\" >-0.09</td>\n",
       "      <td id=\"T_5dc2b_row2_col14\" class=\"data row2 col14\" >-0.09</td>\n",
       "      <td id=\"T_5dc2b_row2_col15\" class=\"data row2 col15\" >-0.09</td>\n",
       "      <td id=\"T_5dc2b_row2_col16\" class=\"data row2 col16\" >-0.11</td>\n",
       "      <td id=\"T_5dc2b_row2_col17\" class=\"data row2 col17\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row2_col18\" class=\"data row2 col18\" >0.03</td>\n",
       "      <td id=\"T_5dc2b_row2_col19\" class=\"data row2 col19\" >0.42</td>\n",
       "      <td id=\"T_5dc2b_row2_col20\" class=\"data row2 col20\" >0.13</td>\n",
       "      <td id=\"T_5dc2b_row2_col21\" class=\"data row2 col21\" >-0.02</td>\n",
       "      <td id=\"T_5dc2b_row2_col22\" class=\"data row2 col22\" >-0.20</td>\n",
       "      <td id=\"T_5dc2b_row2_col23\" class=\"data row2 col23\" >-0.23</td>\n",
       "      <td id=\"T_5dc2b_row2_col24\" class=\"data row2 col24\" >-0.20</td>\n",
       "      <td id=\"T_5dc2b_row2_col25\" class=\"data row2 col25\" >0.14</td>\n",
       "      <td id=\"T_5dc2b_row2_col26\" class=\"data row2 col26\" >-0.19</td>\n",
       "      <td id=\"T_5dc2b_row2_col27\" class=\"data row2 col27\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row2_col28\" class=\"data row2 col28\" >-0.18</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row3\" class=\"row_heading level0 row3\" >min_time_two_events</th>\n",
       "      <td id=\"T_5dc2b_row3_col0\" class=\"data row3 col0\" >-0.09</td>\n",
       "      <td id=\"T_5dc2b_row3_col1\" class=\"data row3 col1\" >-0.07</td>\n",
       "      <td id=\"T_5dc2b_row3_col2\" class=\"data row3 col2\" >0.03</td>\n",
       "      <td id=\"T_5dc2b_row3_col3\" class=\"data row3 col3\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row3_col4\" class=\"data row3 col4\" >0.57</td>\n",
       "      <td id=\"T_5dc2b_row3_col5\" class=\"data row3 col5\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row3_col6\" class=\"data row3 col6\" >0.59</td>\n",
       "      <td id=\"T_5dc2b_row3_col7\" class=\"data row3 col7\" >0.91</td>\n",
       "      <td id=\"T_5dc2b_row3_col8\" class=\"data row3 col8\" >0.51</td>\n",
       "      <td id=\"T_5dc2b_row3_col9\" class=\"data row3 col9\" >0.40</td>\n",
       "      <td id=\"T_5dc2b_row3_col10\" class=\"data row3 col10\" >0.19</td>\n",
       "      <td id=\"T_5dc2b_row3_col11\" class=\"data row3 col11\" >0.37</td>\n",
       "      <td id=\"T_5dc2b_row3_col12\" class=\"data row3 col12\" >0.34</td>\n",
       "      <td id=\"T_5dc2b_row3_col13\" class=\"data row3 col13\" >0.37</td>\n",
       "      <td id=\"T_5dc2b_row3_col14\" class=\"data row3 col14\" >0.34</td>\n",
       "      <td id=\"T_5dc2b_row3_col15\" class=\"data row3 col15\" >0.36</td>\n",
       "      <td id=\"T_5dc2b_row3_col16\" class=\"data row3 col16\" >0.31</td>\n",
       "      <td id=\"T_5dc2b_row3_col17\" class=\"data row3 col17\" >0.26</td>\n",
       "      <td id=\"T_5dc2b_row3_col18\" class=\"data row3 col18\" >0.18</td>\n",
       "      <td id=\"T_5dc2b_row3_col19\" class=\"data row3 col19\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row3_col20\" class=\"data row3 col20\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row3_col21\" class=\"data row3 col21\" >-0.11</td>\n",
       "      <td id=\"T_5dc2b_row3_col22\" class=\"data row3 col22\" >0.09</td>\n",
       "      <td id=\"T_5dc2b_row3_col23\" class=\"data row3 col23\" >0.05</td>\n",
       "      <td id=\"T_5dc2b_row3_col24\" class=\"data row3 col24\" >0.08</td>\n",
       "      <td id=\"T_5dc2b_row3_col25\" class=\"data row3 col25\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row3_col26\" class=\"data row3 col26\" >0.02</td>\n",
       "      <td id=\"T_5dc2b_row3_col27\" class=\"data row3 col27\" >0.01</td>\n",
       "      <td id=\"T_5dc2b_row3_col28\" class=\"data row3 col28\" >0.01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row4\" class=\"row_heading level0 row4\" >mean_time_two_events</th>\n",
       "      <td id=\"T_5dc2b_row4_col0\" class=\"data row4 col0\" >-0.27</td>\n",
       "      <td id=\"T_5dc2b_row4_col1\" class=\"data row4 col1\" >-0.11</td>\n",
       "      <td id=\"T_5dc2b_row4_col2\" class=\"data row4 col2\" >-0.01</td>\n",
       "      <td id=\"T_5dc2b_row4_col3\" class=\"data row4 col3\" >0.57</td>\n",
       "      <td id=\"T_5dc2b_row4_col4\" class=\"data row4 col4\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row4_col5\" class=\"data row4 col5\" >0.59</td>\n",
       "      <td id=\"T_5dc2b_row4_col6\" class=\"data row4 col6\" >0.83</td>\n",
       "      <td id=\"T_5dc2b_row4_col7\" class=\"data row4 col7\" >0.71</td>\n",
       "      <td id=\"T_5dc2b_row4_col8\" class=\"data row4 col8\" >0.98</td>\n",
       "      <td id=\"T_5dc2b_row4_col9\" class=\"data row4 col9\" >0.94</td>\n",
       "      <td id=\"T_5dc2b_row4_col10\" class=\"data row4 col10\" >0.64</td>\n",
       "      <td id=\"T_5dc2b_row4_col11\" class=\"data row4 col11\" >0.61</td>\n",
       "      <td id=\"T_5dc2b_row4_col12\" class=\"data row4 col12\" >0.71</td>\n",
       "      <td id=\"T_5dc2b_row4_col13\" class=\"data row4 col13\" >0.61</td>\n",
       "      <td id=\"T_5dc2b_row4_col14\" class=\"data row4 col14\" >0.70</td>\n",
       "      <td id=\"T_5dc2b_row4_col15\" class=\"data row4 col15\" >0.64</td>\n",
       "      <td id=\"T_5dc2b_row4_col16\" class=\"data row4 col16\" >0.70</td>\n",
       "      <td id=\"T_5dc2b_row4_col17\" class=\"data row4 col17\" >0.67</td>\n",
       "      <td id=\"T_5dc2b_row4_col18\" class=\"data row4 col18\" >0.53</td>\n",
       "      <td id=\"T_5dc2b_row4_col19\" class=\"data row4 col19\" >-0.27</td>\n",
       "      <td id=\"T_5dc2b_row4_col20\" class=\"data row4 col20\" >-0.24</td>\n",
       "      <td id=\"T_5dc2b_row4_col21\" class=\"data row4 col21\" >-0.24</td>\n",
       "      <td id=\"T_5dc2b_row4_col22\" class=\"data row4 col22\" >0.23</td>\n",
       "      <td id=\"T_5dc2b_row4_col23\" class=\"data row4 col23\" >0.31</td>\n",
       "      <td id=\"T_5dc2b_row4_col24\" class=\"data row4 col24\" >0.28</td>\n",
       "      <td id=\"T_5dc2b_row4_col25\" class=\"data row4 col25\" >-0.30</td>\n",
       "      <td id=\"T_5dc2b_row4_col26\" class=\"data row4 col26\" >-0.03</td>\n",
       "      <td id=\"T_5dc2b_row4_col27\" class=\"data row4 col27\" >-0.02</td>\n",
       "      <td id=\"T_5dc2b_row4_col28\" class=\"data row4 col28\" >0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row5\" class=\"row_heading level0 row5\" >10_p_time_two_events</th>\n",
       "      <td id=\"T_5dc2b_row5_col0\" class=\"data row5 col0\" >-0.09</td>\n",
       "      <td id=\"T_5dc2b_row5_col1\" class=\"data row5 col1\" >-0.07</td>\n",
       "      <td id=\"T_5dc2b_row5_col2\" class=\"data row5 col2\" >0.02</td>\n",
       "      <td id=\"T_5dc2b_row5_col3\" class=\"data row5 col3\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row5_col4\" class=\"data row5 col4\" >0.59</td>\n",
       "      <td id=\"T_5dc2b_row5_col5\" class=\"data row5 col5\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row5_col6\" class=\"data row5 col6\" >0.64</td>\n",
       "      <td id=\"T_5dc2b_row5_col7\" class=\"data row5 col7\" >0.93</td>\n",
       "      <td id=\"T_5dc2b_row5_col8\" class=\"data row5 col8\" >0.54</td>\n",
       "      <td id=\"T_5dc2b_row5_col9\" class=\"data row5 col9\" >0.42</td>\n",
       "      <td id=\"T_5dc2b_row5_col10\" class=\"data row5 col10\" >0.20</td>\n",
       "      <td id=\"T_5dc2b_row5_col11\" class=\"data row5 col11\" >0.37</td>\n",
       "      <td id=\"T_5dc2b_row5_col12\" class=\"data row5 col12\" >0.35</td>\n",
       "      <td id=\"T_5dc2b_row5_col13\" class=\"data row5 col13\" >0.37</td>\n",
       "      <td id=\"T_5dc2b_row5_col14\" class=\"data row5 col14\" >0.35</td>\n",
       "      <td id=\"T_5dc2b_row5_col15\" class=\"data row5 col15\" >0.36</td>\n",
       "      <td id=\"T_5dc2b_row5_col16\" class=\"data row5 col16\" >0.32</td>\n",
       "      <td id=\"T_5dc2b_row5_col17\" class=\"data row5 col17\" >0.27</td>\n",
       "      <td id=\"T_5dc2b_row5_col18\" class=\"data row5 col18\" >0.19</td>\n",
       "      <td id=\"T_5dc2b_row5_col19\" class=\"data row5 col19\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row5_col20\" class=\"data row5 col20\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row5_col21\" class=\"data row5 col21\" >-0.11</td>\n",
       "      <td id=\"T_5dc2b_row5_col22\" class=\"data row5 col22\" >0.09</td>\n",
       "      <td id=\"T_5dc2b_row5_col23\" class=\"data row5 col23\" >0.06</td>\n",
       "      <td id=\"T_5dc2b_row5_col24\" class=\"data row5 col24\" >0.08</td>\n",
       "      <td id=\"T_5dc2b_row5_col25\" class=\"data row5 col25\" >-0.11</td>\n",
       "      <td id=\"T_5dc2b_row5_col26\" class=\"data row5 col26\" >0.02</td>\n",
       "      <td id=\"T_5dc2b_row5_col27\" class=\"data row5 col27\" >0.02</td>\n",
       "      <td id=\"T_5dc2b_row5_col28\" class=\"data row5 col28\" >0.01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row6\" class=\"row_heading level0 row6\" >med_time_two_events</th>\n",
       "      <td id=\"T_5dc2b_row6_col0\" class=\"data row6 col0\" >-0.13</td>\n",
       "      <td id=\"T_5dc2b_row6_col1\" class=\"data row6 col1\" >-0.07</td>\n",
       "      <td id=\"T_5dc2b_row6_col2\" class=\"data row6 col2\" >-0.04</td>\n",
       "      <td id=\"T_5dc2b_row6_col3\" class=\"data row6 col3\" >0.59</td>\n",
       "      <td id=\"T_5dc2b_row6_col4\" class=\"data row6 col4\" >0.83</td>\n",
       "      <td id=\"T_5dc2b_row6_col5\" class=\"data row6 col5\" >0.64</td>\n",
       "      <td id=\"T_5dc2b_row6_col6\" class=\"data row6 col6\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row6_col7\" class=\"data row6 col7\" >0.85</td>\n",
       "      <td id=\"T_5dc2b_row6_col8\" class=\"data row6 col8\" >0.83</td>\n",
       "      <td id=\"T_5dc2b_row6_col9\" class=\"data row6 col9\" >0.66</td>\n",
       "      <td id=\"T_5dc2b_row6_col10\" class=\"data row6 col10\" >0.37</td>\n",
       "      <td id=\"T_5dc2b_row6_col11\" class=\"data row6 col11\" >0.33</td>\n",
       "      <td id=\"T_5dc2b_row6_col12\" class=\"data row6 col12\" >0.39</td>\n",
       "      <td id=\"T_5dc2b_row6_col13\" class=\"data row6 col13\" >0.33</td>\n",
       "      <td id=\"T_5dc2b_row6_col14\" class=\"data row6 col14\" >0.37</td>\n",
       "      <td id=\"T_5dc2b_row6_col15\" class=\"data row6 col15\" >0.34</td>\n",
       "      <td id=\"T_5dc2b_row6_col16\" class=\"data row6 col16\" >0.39</td>\n",
       "      <td id=\"T_5dc2b_row6_col17\" class=\"data row6 col17\" >0.38</td>\n",
       "      <td id=\"T_5dc2b_row6_col18\" class=\"data row6 col18\" >0.28</td>\n",
       "      <td id=\"T_5dc2b_row6_col19\" class=\"data row6 col19\" >-0.16</td>\n",
       "      <td id=\"T_5dc2b_row6_col20\" class=\"data row6 col20\" >-0.17</td>\n",
       "      <td id=\"T_5dc2b_row6_col21\" class=\"data row6 col21\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row6_col22\" class=\"data row6 col22\" >0.11</td>\n",
       "      <td id=\"T_5dc2b_row6_col23\" class=\"data row6 col23\" >0.16</td>\n",
       "      <td id=\"T_5dc2b_row6_col24\" class=\"data row6 col24\" >0.13</td>\n",
       "      <td id=\"T_5dc2b_row6_col25\" class=\"data row6 col25\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row6_col26\" class=\"data row6 col26\" >0.03</td>\n",
       "      <td id=\"T_5dc2b_row6_col27\" class=\"data row6 col27\" >0.04</td>\n",
       "      <td id=\"T_5dc2b_row6_col28\" class=\"data row6 col28\" >0.05</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row7\" class=\"row_heading level0 row7\" >25_p_time_two_events</th>\n",
       "      <td id=\"T_5dc2b_row7_col0\" class=\"data row7 col0\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row7_col1\" class=\"data row7 col1\" >-0.07</td>\n",
       "      <td id=\"T_5dc2b_row7_col2\" class=\"data row7 col2\" >-0.00</td>\n",
       "      <td id=\"T_5dc2b_row7_col3\" class=\"data row7 col3\" >0.91</td>\n",
       "      <td id=\"T_5dc2b_row7_col4\" class=\"data row7 col4\" >0.71</td>\n",
       "      <td id=\"T_5dc2b_row7_col5\" class=\"data row7 col5\" >0.93</td>\n",
       "      <td id=\"T_5dc2b_row7_col6\" class=\"data row7 col6\" >0.85</td>\n",
       "      <td id=\"T_5dc2b_row7_col7\" class=\"data row7 col7\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row7_col8\" class=\"data row7 col8\" >0.68</td>\n",
       "      <td id=\"T_5dc2b_row7_col9\" class=\"data row7 col9\" >0.52</td>\n",
       "      <td id=\"T_5dc2b_row7_col10\" class=\"data row7 col10\" >0.26</td>\n",
       "      <td id=\"T_5dc2b_row7_col11\" class=\"data row7 col11\" >0.35</td>\n",
       "      <td id=\"T_5dc2b_row7_col12\" class=\"data row7 col12\" >0.35</td>\n",
       "      <td id=\"T_5dc2b_row7_col13\" class=\"data row7 col13\" >0.35</td>\n",
       "      <td id=\"T_5dc2b_row7_col14\" class=\"data row7 col14\" >0.34</td>\n",
       "      <td id=\"T_5dc2b_row7_col15\" class=\"data row7 col15\" >0.35</td>\n",
       "      <td id=\"T_5dc2b_row7_col16\" class=\"data row7 col16\" >0.34</td>\n",
       "      <td id=\"T_5dc2b_row7_col17\" class=\"data row7 col17\" >0.31</td>\n",
       "      <td id=\"T_5dc2b_row7_col18\" class=\"data row7 col18\" >0.22</td>\n",
       "      <td id=\"T_5dc2b_row7_col19\" class=\"data row7 col19\" >-0.15</td>\n",
       "      <td id=\"T_5dc2b_row7_col20\" class=\"data row7 col20\" >-0.13</td>\n",
       "      <td id=\"T_5dc2b_row7_col21\" class=\"data row7 col21\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row7_col22\" class=\"data row7 col22\" >0.09</td>\n",
       "      <td id=\"T_5dc2b_row7_col23\" class=\"data row7 col23\" >0.08</td>\n",
       "      <td id=\"T_5dc2b_row7_col24\" class=\"data row7 col24\" >0.09</td>\n",
       "      <td id=\"T_5dc2b_row7_col25\" class=\"data row7 col25\" >-0.13</td>\n",
       "      <td id=\"T_5dc2b_row7_col26\" class=\"data row7 col26\" >0.04</td>\n",
       "      <td id=\"T_5dc2b_row7_col27\" class=\"data row7 col27\" >0.04</td>\n",
       "      <td id=\"T_5dc2b_row7_col28\" class=\"data row7 col28\" >0.04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row8\" class=\"row_heading level0 row8\" >75_p_time_two_events</th>\n",
       "      <td id=\"T_5dc2b_row8_col0\" class=\"data row8 col0\" >-0.24</td>\n",
       "      <td id=\"T_5dc2b_row8_col1\" class=\"data row8 col1\" >-0.12</td>\n",
       "      <td id=\"T_5dc2b_row8_col2\" class=\"data row8 col2\" >-0.00</td>\n",
       "      <td id=\"T_5dc2b_row8_col3\" class=\"data row8 col3\" >0.51</td>\n",
       "      <td id=\"T_5dc2b_row8_col4\" class=\"data row8 col4\" >0.98</td>\n",
       "      <td id=\"T_5dc2b_row8_col5\" class=\"data row8 col5\" >0.54</td>\n",
       "      <td id=\"T_5dc2b_row8_col6\" class=\"data row8 col6\" >0.83</td>\n",
       "      <td id=\"T_5dc2b_row8_col7\" class=\"data row8 col7\" >0.68</td>\n",
       "      <td id=\"T_5dc2b_row8_col8\" class=\"data row8 col8\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row8_col9\" class=\"data row8 col9\" >0.92</td>\n",
       "      <td id=\"T_5dc2b_row8_col10\" class=\"data row8 col10\" >0.62</td>\n",
       "      <td id=\"T_5dc2b_row8_col11\" class=\"data row8 col11\" >0.59</td>\n",
       "      <td id=\"T_5dc2b_row8_col12\" class=\"data row8 col12\" >0.66</td>\n",
       "      <td id=\"T_5dc2b_row8_col13\" class=\"data row8 col13\" >0.60</td>\n",
       "      <td id=\"T_5dc2b_row8_col14\" class=\"data row8 col14\" >0.66</td>\n",
       "      <td id=\"T_5dc2b_row8_col15\" class=\"data row8 col15\" >0.62</td>\n",
       "      <td id=\"T_5dc2b_row8_col16\" class=\"data row8 col16\" >0.65</td>\n",
       "      <td id=\"T_5dc2b_row8_col17\" class=\"data row8 col17\" >0.62</td>\n",
       "      <td id=\"T_5dc2b_row8_col18\" class=\"data row8 col18\" >0.53</td>\n",
       "      <td id=\"T_5dc2b_row8_col19\" class=\"data row8 col19\" >-0.22</td>\n",
       "      <td id=\"T_5dc2b_row8_col20\" class=\"data row8 col20\" >-0.15</td>\n",
       "      <td id=\"T_5dc2b_row8_col21\" class=\"data row8 col21\" >-0.24</td>\n",
       "      <td id=\"T_5dc2b_row8_col22\" class=\"data row8 col22\" >0.24</td>\n",
       "      <td id=\"T_5dc2b_row8_col23\" class=\"data row8 col23\" >0.30</td>\n",
       "      <td id=\"T_5dc2b_row8_col24\" class=\"data row8 col24\" >0.28</td>\n",
       "      <td id=\"T_5dc2b_row8_col25\" class=\"data row8 col25\" >-0.22</td>\n",
       "      <td id=\"T_5dc2b_row8_col26\" class=\"data row8 col26\" >-0.05</td>\n",
       "      <td id=\"T_5dc2b_row8_col27\" class=\"data row8 col27\" >-0.04</td>\n",
       "      <td id=\"T_5dc2b_row8_col28\" class=\"data row8 col28\" >-0.03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row9\" class=\"row_heading level0 row9\" >90_p_time_two_events</th>\n",
       "      <td id=\"T_5dc2b_row9_col0\" class=\"data row9 col0\" >-0.30</td>\n",
       "      <td id=\"T_5dc2b_row9_col1\" class=\"data row9 col1\" >-0.13</td>\n",
       "      <td id=\"T_5dc2b_row9_col2\" class=\"data row9 col2\" >0.02</td>\n",
       "      <td id=\"T_5dc2b_row9_col3\" class=\"data row9 col3\" >0.40</td>\n",
       "      <td id=\"T_5dc2b_row9_col4\" class=\"data row9 col4\" >0.94</td>\n",
       "      <td id=\"T_5dc2b_row9_col5\" class=\"data row9 col5\" >0.42</td>\n",
       "      <td id=\"T_5dc2b_row9_col6\" class=\"data row9 col6\" >0.66</td>\n",
       "      <td id=\"T_5dc2b_row9_col7\" class=\"data row9 col7\" >0.52</td>\n",
       "      <td id=\"T_5dc2b_row9_col8\" class=\"data row9 col8\" >0.92</td>\n",
       "      <td id=\"T_5dc2b_row9_col9\" class=\"data row9 col9\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row9_col10\" class=\"data row9 col10\" >0.79</td>\n",
       "      <td id=\"T_5dc2b_row9_col11\" class=\"data row9 col11\" >0.57</td>\n",
       "      <td id=\"T_5dc2b_row9_col12\" class=\"data row9 col12\" >0.73</td>\n",
       "      <td id=\"T_5dc2b_row9_col13\" class=\"data row9 col13\" >0.58</td>\n",
       "      <td id=\"T_5dc2b_row9_col14\" class=\"data row9 col14\" >0.71</td>\n",
       "      <td id=\"T_5dc2b_row9_col15\" class=\"data row9 col15\" >0.62</td>\n",
       "      <td id=\"T_5dc2b_row9_col16\" class=\"data row9 col16\" >0.75</td>\n",
       "      <td id=\"T_5dc2b_row9_col17\" class=\"data row9 col17\" >0.75</td>\n",
       "      <td id=\"T_5dc2b_row9_col18\" class=\"data row9 col18\" >0.68</td>\n",
       "      <td id=\"T_5dc2b_row9_col19\" class=\"data row9 col19\" >-0.25</td>\n",
       "      <td id=\"T_5dc2b_row9_col20\" class=\"data row9 col20\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row9_col21\" class=\"data row9 col21\" >-0.30</td>\n",
       "      <td id=\"T_5dc2b_row9_col22\" class=\"data row9 col22\" >0.17</td>\n",
       "      <td id=\"T_5dc2b_row9_col23\" class=\"data row9 col23\" >0.27</td>\n",
       "      <td id=\"T_5dc2b_row9_col24\" class=\"data row9 col24\" >0.24</td>\n",
       "      <td id=\"T_5dc2b_row9_col25\" class=\"data row9 col25\" >-0.24</td>\n",
       "      <td id=\"T_5dc2b_row9_col26\" class=\"data row9 col26\" >-0.08</td>\n",
       "      <td id=\"T_5dc2b_row9_col27\" class=\"data row9 col27\" >-0.06</td>\n",
       "      <td id=\"T_5dc2b_row9_col28\" class=\"data row9 col28\" >-0.05</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row10\" class=\"row_heading level0 row10\" >max_time_two_events</th>\n",
       "      <td id=\"T_5dc2b_row10_col0\" class=\"data row10 col0\" >-0.26</td>\n",
       "      <td id=\"T_5dc2b_row10_col1\" class=\"data row10 col1\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row10_col2\" class=\"data row10 col2\" >0.16</td>\n",
       "      <td id=\"T_5dc2b_row10_col3\" class=\"data row10 col3\" >0.19</td>\n",
       "      <td id=\"T_5dc2b_row10_col4\" class=\"data row10 col4\" >0.64</td>\n",
       "      <td id=\"T_5dc2b_row10_col5\" class=\"data row10 col5\" >0.20</td>\n",
       "      <td id=\"T_5dc2b_row10_col6\" class=\"data row10 col6\" >0.37</td>\n",
       "      <td id=\"T_5dc2b_row10_col7\" class=\"data row10 col7\" >0.26</td>\n",
       "      <td id=\"T_5dc2b_row10_col8\" class=\"data row10 col8\" >0.62</td>\n",
       "      <td id=\"T_5dc2b_row10_col9\" class=\"data row10 col9\" >0.79</td>\n",
       "      <td id=\"T_5dc2b_row10_col10\" class=\"data row10 col10\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row10_col11\" class=\"data row10 col11\" >0.30</td>\n",
       "      <td id=\"T_5dc2b_row10_col12\" class=\"data row10 col12\" >0.52</td>\n",
       "      <td id=\"T_5dc2b_row10_col13\" class=\"data row10 col13\" >0.31</td>\n",
       "      <td id=\"T_5dc2b_row10_col14\" class=\"data row10 col14\" >0.48</td>\n",
       "      <td id=\"T_5dc2b_row10_col15\" class=\"data row10 col15\" >0.36</td>\n",
       "      <td id=\"T_5dc2b_row10_col16\" class=\"data row10 col16\" >0.59</td>\n",
       "      <td id=\"T_5dc2b_row10_col17\" class=\"data row10 col17\" >0.65</td>\n",
       "      <td id=\"T_5dc2b_row10_col18\" class=\"data row10 col18\" >0.81</td>\n",
       "      <td id=\"T_5dc2b_row10_col19\" class=\"data row10 col19\" >0.06</td>\n",
       "      <td id=\"T_5dc2b_row10_col20\" class=\"data row10 col20\" >0.19</td>\n",
       "      <td id=\"T_5dc2b_row10_col21\" class=\"data row10 col21\" >-0.35</td>\n",
       "      <td id=\"T_5dc2b_row10_col22\" class=\"data row10 col22\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row10_col23\" class=\"data row10 col23\" >-0.04</td>\n",
       "      <td id=\"T_5dc2b_row10_col24\" class=\"data row10 col24\" >-0.04</td>\n",
       "      <td id=\"T_5dc2b_row10_col25\" class=\"data row10 col25\" >0.02</td>\n",
       "      <td id=\"T_5dc2b_row10_col26\" class=\"data row10 col26\" >-0.16</td>\n",
       "      <td id=\"T_5dc2b_row10_col27\" class=\"data row10 col27\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row10_col28\" class=\"data row10 col28\" >-0.14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row11\" class=\"row_heading level0 row11\" >min_lifetime_cancel</th>\n",
       "      <td id=\"T_5dc2b_row11_col0\" class=\"data row11 col0\" >-0.02</td>\n",
       "      <td id=\"T_5dc2b_row11_col1\" class=\"data row11 col1\" >0.13</td>\n",
       "      <td id=\"T_5dc2b_row11_col2\" class=\"data row11 col2\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row11_col3\" class=\"data row11 col3\" >0.37</td>\n",
       "      <td id=\"T_5dc2b_row11_col4\" class=\"data row11 col4\" >0.61</td>\n",
       "      <td id=\"T_5dc2b_row11_col5\" class=\"data row11 col5\" >0.37</td>\n",
       "      <td id=\"T_5dc2b_row11_col6\" class=\"data row11 col6\" >0.33</td>\n",
       "      <td id=\"T_5dc2b_row11_col7\" class=\"data row11 col7\" >0.35</td>\n",
       "      <td id=\"T_5dc2b_row11_col8\" class=\"data row11 col8\" >0.59</td>\n",
       "      <td id=\"T_5dc2b_row11_col9\" class=\"data row11 col9\" >0.57</td>\n",
       "      <td id=\"T_5dc2b_row11_col10\" class=\"data row11 col10\" >0.30</td>\n",
       "      <td id=\"T_5dc2b_row11_col11\" class=\"data row11 col11\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row11_col12\" class=\"data row11 col12\" >0.90</td>\n",
       "      <td id=\"T_5dc2b_row11_col13\" class=\"data row11 col13\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row11_col14\" class=\"data row11 col14\" >0.94</td>\n",
       "      <td id=\"T_5dc2b_row11_col15\" class=\"data row11 col15\" >0.99</td>\n",
       "      <td id=\"T_5dc2b_row11_col16\" class=\"data row11 col16\" >0.79</td>\n",
       "      <td id=\"T_5dc2b_row11_col17\" class=\"data row11 col17\" >0.67</td>\n",
       "      <td id=\"T_5dc2b_row11_col18\" class=\"data row11 col18\" >0.46</td>\n",
       "      <td id=\"T_5dc2b_row11_col19\" class=\"data row11 col19\" >-0.36</td>\n",
       "      <td id=\"T_5dc2b_row11_col20\" class=\"data row11 col20\" >-0.30</td>\n",
       "      <td id=\"T_5dc2b_row11_col21\" class=\"data row11 col21\" >-0.06</td>\n",
       "      <td id=\"T_5dc2b_row11_col22\" class=\"data row11 col22\" >0.45</td>\n",
       "      <td id=\"T_5dc2b_row11_col23\" class=\"data row11 col23\" >0.45</td>\n",
       "      <td id=\"T_5dc2b_row11_col24\" class=\"data row11 col24\" >0.46</td>\n",
       "      <td id=\"T_5dc2b_row11_col25\" class=\"data row11 col25\" >-0.34</td>\n",
       "      <td id=\"T_5dc2b_row11_col26\" class=\"data row11 col26\" >0.17</td>\n",
       "      <td id=\"T_5dc2b_row11_col27\" class=\"data row11 col27\" >0.18</td>\n",
       "      <td id=\"T_5dc2b_row11_col28\" class=\"data row11 col28\" >0.17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row12\" class=\"row_heading level0 row12\" >mean_lifetime_cancel</th>\n",
       "      <td id=\"T_5dc2b_row12_col0\" class=\"data row12 col0\" >-0.01</td>\n",
       "      <td id=\"T_5dc2b_row12_col1\" class=\"data row12 col1\" >0.19</td>\n",
       "      <td id=\"T_5dc2b_row12_col2\" class=\"data row12 col2\" >-0.11</td>\n",
       "      <td id=\"T_5dc2b_row12_col3\" class=\"data row12 col3\" >0.34</td>\n",
       "      <td id=\"T_5dc2b_row12_col4\" class=\"data row12 col4\" >0.71</td>\n",
       "      <td id=\"T_5dc2b_row12_col5\" class=\"data row12 col5\" >0.35</td>\n",
       "      <td id=\"T_5dc2b_row12_col6\" class=\"data row12 col6\" >0.39</td>\n",
       "      <td id=\"T_5dc2b_row12_col7\" class=\"data row12 col7\" >0.35</td>\n",
       "      <td id=\"T_5dc2b_row12_col8\" class=\"data row12 col8\" >0.66</td>\n",
       "      <td id=\"T_5dc2b_row12_col9\" class=\"data row12 col9\" >0.73</td>\n",
       "      <td id=\"T_5dc2b_row12_col10\" class=\"data row12 col10\" >0.52</td>\n",
       "      <td id=\"T_5dc2b_row12_col11\" class=\"data row12 col11\" >0.90</td>\n",
       "      <td id=\"T_5dc2b_row12_col12\" class=\"data row12 col12\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row12_col13\" class=\"data row12 col13\" >0.90</td>\n",
       "      <td id=\"T_5dc2b_row12_col14\" class=\"data row12 col14\" >0.99</td>\n",
       "      <td id=\"T_5dc2b_row12_col15\" class=\"data row12 col15\" >0.94</td>\n",
       "      <td id=\"T_5dc2b_row12_col16\" class=\"data row12 col16\" >0.98</td>\n",
       "      <td id=\"T_5dc2b_row12_col17\" class=\"data row12 col17\" >0.92</td>\n",
       "      <td id=\"T_5dc2b_row12_col18\" class=\"data row12 col18\" >0.72</td>\n",
       "      <td id=\"T_5dc2b_row12_col19\" class=\"data row12 col19\" >-0.39</td>\n",
       "      <td id=\"T_5dc2b_row12_col20\" class=\"data row12 col20\" >-0.31</td>\n",
       "      <td id=\"T_5dc2b_row12_col21\" class=\"data row12 col21\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row12_col22\" class=\"data row12 col22\" >0.35</td>\n",
       "      <td id=\"T_5dc2b_row12_col23\" class=\"data row12 col23\" >0.39</td>\n",
       "      <td id=\"T_5dc2b_row12_col24\" class=\"data row12 col24\" >0.39</td>\n",
       "      <td id=\"T_5dc2b_row12_col25\" class=\"data row12 col25\" >-0.39</td>\n",
       "      <td id=\"T_5dc2b_row12_col26\" class=\"data row12 col26\" >0.15</td>\n",
       "      <td id=\"T_5dc2b_row12_col27\" class=\"data row12 col27\" >0.16</td>\n",
       "      <td id=\"T_5dc2b_row12_col28\" class=\"data row12 col28\" >0.16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row13\" class=\"row_heading level0 row13\" >10_p_lifetime_cancel</th>\n",
       "      <td id=\"T_5dc2b_row13_col0\" class=\"data row13 col0\" >-0.02</td>\n",
       "      <td id=\"T_5dc2b_row13_col1\" class=\"data row13 col1\" >0.13</td>\n",
       "      <td id=\"T_5dc2b_row13_col2\" class=\"data row13 col2\" >-0.09</td>\n",
       "      <td id=\"T_5dc2b_row13_col3\" class=\"data row13 col3\" >0.37</td>\n",
       "      <td id=\"T_5dc2b_row13_col4\" class=\"data row13 col4\" >0.61</td>\n",
       "      <td id=\"T_5dc2b_row13_col5\" class=\"data row13 col5\" >0.37</td>\n",
       "      <td id=\"T_5dc2b_row13_col6\" class=\"data row13 col6\" >0.33</td>\n",
       "      <td id=\"T_5dc2b_row13_col7\" class=\"data row13 col7\" >0.35</td>\n",
       "      <td id=\"T_5dc2b_row13_col8\" class=\"data row13 col8\" >0.60</td>\n",
       "      <td id=\"T_5dc2b_row13_col9\" class=\"data row13 col9\" >0.58</td>\n",
       "      <td id=\"T_5dc2b_row13_col10\" class=\"data row13 col10\" >0.31</td>\n",
       "      <td id=\"T_5dc2b_row13_col11\" class=\"data row13 col11\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row13_col12\" class=\"data row13 col12\" >0.90</td>\n",
       "      <td id=\"T_5dc2b_row13_col13\" class=\"data row13 col13\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row13_col14\" class=\"data row13 col14\" >0.94</td>\n",
       "      <td id=\"T_5dc2b_row13_col15\" class=\"data row13 col15\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row13_col16\" class=\"data row13 col16\" >0.80</td>\n",
       "      <td id=\"T_5dc2b_row13_col17\" class=\"data row13 col17\" >0.68</td>\n",
       "      <td id=\"T_5dc2b_row13_col18\" class=\"data row13 col18\" >0.47</td>\n",
       "      <td id=\"T_5dc2b_row13_col19\" class=\"data row13 col19\" >-0.36</td>\n",
       "      <td id=\"T_5dc2b_row13_col20\" class=\"data row13 col20\" >-0.29</td>\n",
       "      <td id=\"T_5dc2b_row13_col21\" class=\"data row13 col21\" >-0.07</td>\n",
       "      <td id=\"T_5dc2b_row13_col22\" class=\"data row13 col22\" >0.44</td>\n",
       "      <td id=\"T_5dc2b_row13_col23\" class=\"data row13 col23\" >0.45</td>\n",
       "      <td id=\"T_5dc2b_row13_col24\" class=\"data row13 col24\" >0.46</td>\n",
       "      <td id=\"T_5dc2b_row13_col25\" class=\"data row13 col25\" >-0.34</td>\n",
       "      <td id=\"T_5dc2b_row13_col26\" class=\"data row13 col26\" >0.17</td>\n",
       "      <td id=\"T_5dc2b_row13_col27\" class=\"data row13 col27\" >0.17</td>\n",
       "      <td id=\"T_5dc2b_row13_col28\" class=\"data row13 col28\" >0.17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row14\" class=\"row_heading level0 row14\" >med_lifetime_cancel</th>\n",
       "      <td id=\"T_5dc2b_row14_col0\" class=\"data row14 col0\" >-0.02</td>\n",
       "      <td id=\"T_5dc2b_row14_col1\" class=\"data row14 col1\" >0.18</td>\n",
       "      <td id=\"T_5dc2b_row14_col2\" class=\"data row14 col2\" >-0.09</td>\n",
       "      <td id=\"T_5dc2b_row14_col3\" class=\"data row14 col3\" >0.34</td>\n",
       "      <td id=\"T_5dc2b_row14_col4\" class=\"data row14 col4\" >0.70</td>\n",
       "      <td id=\"T_5dc2b_row14_col5\" class=\"data row14 col5\" >0.35</td>\n",
       "      <td id=\"T_5dc2b_row14_col6\" class=\"data row14 col6\" >0.37</td>\n",
       "      <td id=\"T_5dc2b_row14_col7\" class=\"data row14 col7\" >0.34</td>\n",
       "      <td id=\"T_5dc2b_row14_col8\" class=\"data row14 col8\" >0.66</td>\n",
       "      <td id=\"T_5dc2b_row14_col9\" class=\"data row14 col9\" >0.71</td>\n",
       "      <td id=\"T_5dc2b_row14_col10\" class=\"data row14 col10\" >0.48</td>\n",
       "      <td id=\"T_5dc2b_row14_col11\" class=\"data row14 col11\" >0.94</td>\n",
       "      <td id=\"T_5dc2b_row14_col12\" class=\"data row14 col12\" >0.99</td>\n",
       "      <td id=\"T_5dc2b_row14_col13\" class=\"data row14 col13\" >0.94</td>\n",
       "      <td id=\"T_5dc2b_row14_col14\" class=\"data row14 col14\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row14_col15\" class=\"data row14 col15\" >0.97</td>\n",
       "      <td id=\"T_5dc2b_row14_col16\" class=\"data row14 col16\" >0.94</td>\n",
       "      <td id=\"T_5dc2b_row14_col17\" class=\"data row14 col17\" >0.87</td>\n",
       "      <td id=\"T_5dc2b_row14_col18\" class=\"data row14 col18\" >0.66</td>\n",
       "      <td id=\"T_5dc2b_row14_col19\" class=\"data row14 col19\" >-0.37</td>\n",
       "      <td id=\"T_5dc2b_row14_col20\" class=\"data row14 col20\" >-0.30</td>\n",
       "      <td id=\"T_5dc2b_row14_col21\" class=\"data row14 col21\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row14_col22\" class=\"data row14 col22\" >0.37</td>\n",
       "      <td id=\"T_5dc2b_row14_col23\" class=\"data row14 col23\" >0.40</td>\n",
       "      <td id=\"T_5dc2b_row14_col24\" class=\"data row14 col24\" >0.40</td>\n",
       "      <td id=\"T_5dc2b_row14_col25\" class=\"data row14 col25\" >-0.37</td>\n",
       "      <td id=\"T_5dc2b_row14_col26\" class=\"data row14 col26\" >0.14</td>\n",
       "      <td id=\"T_5dc2b_row14_col27\" class=\"data row14 col27\" >0.15</td>\n",
       "      <td id=\"T_5dc2b_row14_col28\" class=\"data row14 col28\" >0.15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row15\" class=\"row_heading level0 row15\" >25_p_lifetime_cancel</th>\n",
       "      <td id=\"T_5dc2b_row15_col0\" class=\"data row15 col0\" >-0.03</td>\n",
       "      <td id=\"T_5dc2b_row15_col1\" class=\"data row15 col1\" >0.13</td>\n",
       "      <td id=\"T_5dc2b_row15_col2\" class=\"data row15 col2\" >-0.09</td>\n",
       "      <td id=\"T_5dc2b_row15_col3\" class=\"data row15 col3\" >0.36</td>\n",
       "      <td id=\"T_5dc2b_row15_col4\" class=\"data row15 col4\" >0.64</td>\n",
       "      <td id=\"T_5dc2b_row15_col5\" class=\"data row15 col5\" >0.36</td>\n",
       "      <td id=\"T_5dc2b_row15_col6\" class=\"data row15 col6\" >0.34</td>\n",
       "      <td id=\"T_5dc2b_row15_col7\" class=\"data row15 col7\" >0.35</td>\n",
       "      <td id=\"T_5dc2b_row15_col8\" class=\"data row15 col8\" >0.62</td>\n",
       "      <td id=\"T_5dc2b_row15_col9\" class=\"data row15 col9\" >0.62</td>\n",
       "      <td id=\"T_5dc2b_row15_col10\" class=\"data row15 col10\" >0.36</td>\n",
       "      <td id=\"T_5dc2b_row15_col11\" class=\"data row15 col11\" >0.99</td>\n",
       "      <td id=\"T_5dc2b_row15_col12\" class=\"data row15 col12\" >0.94</td>\n",
       "      <td id=\"T_5dc2b_row15_col13\" class=\"data row15 col13\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row15_col14\" class=\"data row15 col14\" >0.97</td>\n",
       "      <td id=\"T_5dc2b_row15_col15\" class=\"data row15 col15\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row15_col16\" class=\"data row15 col16\" >0.85</td>\n",
       "      <td id=\"T_5dc2b_row15_col17\" class=\"data row15 col17\" >0.74</td>\n",
       "      <td id=\"T_5dc2b_row15_col18\" class=\"data row15 col18\" >0.53</td>\n",
       "      <td id=\"T_5dc2b_row15_col19\" class=\"data row15 col19\" >-0.36</td>\n",
       "      <td id=\"T_5dc2b_row15_col20\" class=\"data row15 col20\" >-0.29</td>\n",
       "      <td id=\"T_5dc2b_row15_col21\" class=\"data row15 col21\" >-0.09</td>\n",
       "      <td id=\"T_5dc2b_row15_col22\" class=\"data row15 col22\" >0.42</td>\n",
       "      <td id=\"T_5dc2b_row15_col23\" class=\"data row15 col23\" >0.44</td>\n",
       "      <td id=\"T_5dc2b_row15_col24\" class=\"data row15 col24\" >0.45</td>\n",
       "      <td id=\"T_5dc2b_row15_col25\" class=\"data row15 col25\" >-0.34</td>\n",
       "      <td id=\"T_5dc2b_row15_col26\" class=\"data row15 col26\" >0.16</td>\n",
       "      <td id=\"T_5dc2b_row15_col27\" class=\"data row15 col27\" >0.17</td>\n",
       "      <td id=\"T_5dc2b_row15_col28\" class=\"data row15 col28\" >0.16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row16\" class=\"row_heading level0 row16\" >75_p_lifetime_cancel</th>\n",
       "      <td id=\"T_5dc2b_row16_col0\" class=\"data row16 col0\" >0.01</td>\n",
       "      <td id=\"T_5dc2b_row16_col1\" class=\"data row16 col1\" >0.22</td>\n",
       "      <td id=\"T_5dc2b_row16_col2\" class=\"data row16 col2\" >-0.11</td>\n",
       "      <td id=\"T_5dc2b_row16_col3\" class=\"data row16 col3\" >0.31</td>\n",
       "      <td id=\"T_5dc2b_row16_col4\" class=\"data row16 col4\" >0.70</td>\n",
       "      <td id=\"T_5dc2b_row16_col5\" class=\"data row16 col5\" >0.32</td>\n",
       "      <td id=\"T_5dc2b_row16_col6\" class=\"data row16 col6\" >0.39</td>\n",
       "      <td id=\"T_5dc2b_row16_col7\" class=\"data row16 col7\" >0.34</td>\n",
       "      <td id=\"T_5dc2b_row16_col8\" class=\"data row16 col8\" >0.65</td>\n",
       "      <td id=\"T_5dc2b_row16_col9\" class=\"data row16 col9\" >0.75</td>\n",
       "      <td id=\"T_5dc2b_row16_col10\" class=\"data row16 col10\" >0.59</td>\n",
       "      <td id=\"T_5dc2b_row16_col11\" class=\"data row16 col11\" >0.79</td>\n",
       "      <td id=\"T_5dc2b_row16_col12\" class=\"data row16 col12\" >0.98</td>\n",
       "      <td id=\"T_5dc2b_row16_col13\" class=\"data row16 col13\" >0.80</td>\n",
       "      <td id=\"T_5dc2b_row16_col14\" class=\"data row16 col14\" >0.94</td>\n",
       "      <td id=\"T_5dc2b_row16_col15\" class=\"data row16 col15\" >0.85</td>\n",
       "      <td id=\"T_5dc2b_row16_col16\" class=\"data row16 col16\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row16_col17\" class=\"data row16 col17\" >0.98</td>\n",
       "      <td id=\"T_5dc2b_row16_col18\" class=\"data row16 col18\" >0.81</td>\n",
       "      <td id=\"T_5dc2b_row16_col19\" class=\"data row16 col19\" >-0.37</td>\n",
       "      <td id=\"T_5dc2b_row16_col20\" class=\"data row16 col20\" >-0.27</td>\n",
       "      <td id=\"T_5dc2b_row16_col21\" class=\"data row16 col21\" >-0.22</td>\n",
       "      <td id=\"T_5dc2b_row16_col22\" class=\"data row16 col22\" >0.28</td>\n",
       "      <td id=\"T_5dc2b_row16_col23\" class=\"data row16 col23\" >0.34</td>\n",
       "      <td id=\"T_5dc2b_row16_col24\" class=\"data row16 col24\" >0.33</td>\n",
       "      <td id=\"T_5dc2b_row16_col25\" class=\"data row16 col25\" >-0.35</td>\n",
       "      <td id=\"T_5dc2b_row16_col26\" class=\"data row16 col26\" >0.13</td>\n",
       "      <td id=\"T_5dc2b_row16_col27\" class=\"data row16 col27\" >0.14</td>\n",
       "      <td id=\"T_5dc2b_row16_col28\" class=\"data row16 col28\" >0.14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row17\" class=\"row_heading level0 row17\" >90_p_lifetime_cancel</th>\n",
       "      <td id=\"T_5dc2b_row17_col0\" class=\"data row17 col0\" >0.01</td>\n",
       "      <td id=\"T_5dc2b_row17_col1\" class=\"data row17 col1\" >0.20</td>\n",
       "      <td id=\"T_5dc2b_row17_col2\" class=\"data row17 col2\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row17_col3\" class=\"data row17 col3\" >0.26</td>\n",
       "      <td id=\"T_5dc2b_row17_col4\" class=\"data row17 col4\" >0.67</td>\n",
       "      <td id=\"T_5dc2b_row17_col5\" class=\"data row17 col5\" >0.27</td>\n",
       "      <td id=\"T_5dc2b_row17_col6\" class=\"data row17 col6\" >0.38</td>\n",
       "      <td id=\"T_5dc2b_row17_col7\" class=\"data row17 col7\" >0.31</td>\n",
       "      <td id=\"T_5dc2b_row17_col8\" class=\"data row17 col8\" >0.62</td>\n",
       "      <td id=\"T_5dc2b_row17_col9\" class=\"data row17 col9\" >0.75</td>\n",
       "      <td id=\"T_5dc2b_row17_col10\" class=\"data row17 col10\" >0.65</td>\n",
       "      <td id=\"T_5dc2b_row17_col11\" class=\"data row17 col11\" >0.67</td>\n",
       "      <td id=\"T_5dc2b_row17_col12\" class=\"data row17 col12\" >0.92</td>\n",
       "      <td id=\"T_5dc2b_row17_col13\" class=\"data row17 col13\" >0.68</td>\n",
       "      <td id=\"T_5dc2b_row17_col14\" class=\"data row17 col14\" >0.87</td>\n",
       "      <td id=\"T_5dc2b_row17_col15\" class=\"data row17 col15\" >0.74</td>\n",
       "      <td id=\"T_5dc2b_row17_col16\" class=\"data row17 col16\" >0.98</td>\n",
       "      <td id=\"T_5dc2b_row17_col17\" class=\"data row17 col17\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row17_col18\" class=\"data row17 col18\" >0.87</td>\n",
       "      <td id=\"T_5dc2b_row17_col19\" class=\"data row17 col19\" >-0.33</td>\n",
       "      <td id=\"T_5dc2b_row17_col20\" class=\"data row17 col20\" >-0.23</td>\n",
       "      <td id=\"T_5dc2b_row17_col21\" class=\"data row17 col21\" >-0.27</td>\n",
       "      <td id=\"T_5dc2b_row17_col22\" class=\"data row17 col22\" >0.21</td>\n",
       "      <td id=\"T_5dc2b_row17_col23\" class=\"data row17 col23\" >0.27</td>\n",
       "      <td id=\"T_5dc2b_row17_col24\" class=\"data row17 col24\" >0.26</td>\n",
       "      <td id=\"T_5dc2b_row17_col25\" class=\"data row17 col25\" >-0.31</td>\n",
       "      <td id=\"T_5dc2b_row17_col26\" class=\"data row17 col26\" >0.10</td>\n",
       "      <td id=\"T_5dc2b_row17_col27\" class=\"data row17 col27\" >0.12</td>\n",
       "      <td id=\"T_5dc2b_row17_col28\" class=\"data row17 col28\" >0.12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row18\" class=\"row_heading level0 row18\" >max_lifetime_cancel</th>\n",
       "      <td id=\"T_5dc2b_row18_col0\" class=\"data row18 col0\" >0.01</td>\n",
       "      <td id=\"T_5dc2b_row18_col1\" class=\"data row18 col1\" >0.10</td>\n",
       "      <td id=\"T_5dc2b_row18_col2\" class=\"data row18 col2\" >0.03</td>\n",
       "      <td id=\"T_5dc2b_row18_col3\" class=\"data row18 col3\" >0.18</td>\n",
       "      <td id=\"T_5dc2b_row18_col4\" class=\"data row18 col4\" >0.53</td>\n",
       "      <td id=\"T_5dc2b_row18_col5\" class=\"data row18 col5\" >0.19</td>\n",
       "      <td id=\"T_5dc2b_row18_col6\" class=\"data row18 col6\" >0.28</td>\n",
       "      <td id=\"T_5dc2b_row18_col7\" class=\"data row18 col7\" >0.22</td>\n",
       "      <td id=\"T_5dc2b_row18_col8\" class=\"data row18 col8\" >0.53</td>\n",
       "      <td id=\"T_5dc2b_row18_col9\" class=\"data row18 col9\" >0.68</td>\n",
       "      <td id=\"T_5dc2b_row18_col10\" class=\"data row18 col10\" >0.81</td>\n",
       "      <td id=\"T_5dc2b_row18_col11\" class=\"data row18 col11\" >0.46</td>\n",
       "      <td id=\"T_5dc2b_row18_col12\" class=\"data row18 col12\" >0.72</td>\n",
       "      <td id=\"T_5dc2b_row18_col13\" class=\"data row18 col13\" >0.47</td>\n",
       "      <td id=\"T_5dc2b_row18_col14\" class=\"data row18 col14\" >0.66</td>\n",
       "      <td id=\"T_5dc2b_row18_col15\" class=\"data row18 col15\" >0.53</td>\n",
       "      <td id=\"T_5dc2b_row18_col16\" class=\"data row18 col16\" >0.81</td>\n",
       "      <td id=\"T_5dc2b_row18_col17\" class=\"data row18 col17\" >0.87</td>\n",
       "      <td id=\"T_5dc2b_row18_col18\" class=\"data row18 col18\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row18_col19\" class=\"data row18 col19\" >-0.07</td>\n",
       "      <td id=\"T_5dc2b_row18_col20\" class=\"data row18 col20\" >0.15</td>\n",
       "      <td id=\"T_5dc2b_row18_col21\" class=\"data row18 col21\" >-0.36</td>\n",
       "      <td id=\"T_5dc2b_row18_col22\" class=\"data row18 col22\" >0.03</td>\n",
       "      <td id=\"T_5dc2b_row18_col23\" class=\"data row18 col23\" >0.05</td>\n",
       "      <td id=\"T_5dc2b_row18_col24\" class=\"data row18 col24\" >0.08</td>\n",
       "      <td id=\"T_5dc2b_row18_col25\" class=\"data row18 col25\" >0.01</td>\n",
       "      <td id=\"T_5dc2b_row18_col26\" class=\"data row18 col26\" >-0.00</td>\n",
       "      <td id=\"T_5dc2b_row18_col27\" class=\"data row18 col27\" >0.01</td>\n",
       "      <td id=\"T_5dc2b_row18_col28\" class=\"data row18 col28\" >-0.01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row19\" class=\"row_heading level0 row19\" >NbTradeVenueMic</th>\n",
       "      <td id=\"T_5dc2b_row19_col0\" class=\"data row19 col0\" >0.06</td>\n",
       "      <td id=\"T_5dc2b_row19_col1\" class=\"data row19 col1\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row19_col2\" class=\"data row19 col2\" >0.42</td>\n",
       "      <td id=\"T_5dc2b_row19_col3\" class=\"data row19 col3\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row19_col4\" class=\"data row19 col4\" >-0.27</td>\n",
       "      <td id=\"T_5dc2b_row19_col5\" class=\"data row19 col5\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row19_col6\" class=\"data row19 col6\" >-0.16</td>\n",
       "      <td id=\"T_5dc2b_row19_col7\" class=\"data row19 col7\" >-0.15</td>\n",
       "      <td id=\"T_5dc2b_row19_col8\" class=\"data row19 col8\" >-0.22</td>\n",
       "      <td id=\"T_5dc2b_row19_col9\" class=\"data row19 col9\" >-0.25</td>\n",
       "      <td id=\"T_5dc2b_row19_col10\" class=\"data row19 col10\" >0.06</td>\n",
       "      <td id=\"T_5dc2b_row19_col11\" class=\"data row19 col11\" >-0.36</td>\n",
       "      <td id=\"T_5dc2b_row19_col12\" class=\"data row19 col12\" >-0.39</td>\n",
       "      <td id=\"T_5dc2b_row19_col13\" class=\"data row19 col13\" >-0.36</td>\n",
       "      <td id=\"T_5dc2b_row19_col14\" class=\"data row19 col14\" >-0.37</td>\n",
       "      <td id=\"T_5dc2b_row19_col15\" class=\"data row19 col15\" >-0.36</td>\n",
       "      <td id=\"T_5dc2b_row19_col16\" class=\"data row19 col16\" >-0.37</td>\n",
       "      <td id=\"T_5dc2b_row19_col17\" class=\"data row19 col17\" >-0.33</td>\n",
       "      <td id=\"T_5dc2b_row19_col18\" class=\"data row19 col18\" >-0.07</td>\n",
       "      <td id=\"T_5dc2b_row19_col19\" class=\"data row19 col19\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row19_col20\" class=\"data row19 col20\" >0.41</td>\n",
       "      <td id=\"T_5dc2b_row19_col21\" class=\"data row19 col21\" >-0.16</td>\n",
       "      <td id=\"T_5dc2b_row19_col22\" class=\"data row19 col22\" >-0.35</td>\n",
       "      <td id=\"T_5dc2b_row19_col23\" class=\"data row19 col23\" >-0.50</td>\n",
       "      <td id=\"T_5dc2b_row19_col24\" class=\"data row19 col24\" >-0.40</td>\n",
       "      <td id=\"T_5dc2b_row19_col25\" class=\"data row19 col25\" >0.63</td>\n",
       "      <td id=\"T_5dc2b_row19_col26\" class=\"data row19 col26\" >-0.20</td>\n",
       "      <td id=\"T_5dc2b_row19_col27\" class=\"data row19 col27\" >-0.21</td>\n",
       "      <td id=\"T_5dc2b_row19_col28\" class=\"data row19 col28\" >-0.23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row20\" class=\"row_heading level0 row20\" >MaxNbTradesBySecond</th>\n",
       "      <td id=\"T_5dc2b_row20_col0\" class=\"data row20 col0\" >-0.07</td>\n",
       "      <td id=\"T_5dc2b_row20_col1\" class=\"data row20 col1\" >-0.13</td>\n",
       "      <td id=\"T_5dc2b_row20_col2\" class=\"data row20 col2\" >0.13</td>\n",
       "      <td id=\"T_5dc2b_row20_col3\" class=\"data row20 col3\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row20_col4\" class=\"data row20 col4\" >-0.24</td>\n",
       "      <td id=\"T_5dc2b_row20_col5\" class=\"data row20 col5\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row20_col6\" class=\"data row20 col6\" >-0.17</td>\n",
       "      <td id=\"T_5dc2b_row20_col7\" class=\"data row20 col7\" >-0.13</td>\n",
       "      <td id=\"T_5dc2b_row20_col8\" class=\"data row20 col8\" >-0.15</td>\n",
       "      <td id=\"T_5dc2b_row20_col9\" class=\"data row20 col9\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row20_col10\" class=\"data row20 col10\" >0.19</td>\n",
       "      <td id=\"T_5dc2b_row20_col11\" class=\"data row20 col11\" >-0.30</td>\n",
       "      <td id=\"T_5dc2b_row20_col12\" class=\"data row20 col12\" >-0.31</td>\n",
       "      <td id=\"T_5dc2b_row20_col13\" class=\"data row20 col13\" >-0.29</td>\n",
       "      <td id=\"T_5dc2b_row20_col14\" class=\"data row20 col14\" >-0.30</td>\n",
       "      <td id=\"T_5dc2b_row20_col15\" class=\"data row20 col15\" >-0.29</td>\n",
       "      <td id=\"T_5dc2b_row20_col16\" class=\"data row20 col16\" >-0.27</td>\n",
       "      <td id=\"T_5dc2b_row20_col17\" class=\"data row20 col17\" >-0.23</td>\n",
       "      <td id=\"T_5dc2b_row20_col18\" class=\"data row20 col18\" >0.15</td>\n",
       "      <td id=\"T_5dc2b_row20_col19\" class=\"data row20 col19\" >0.41</td>\n",
       "      <td id=\"T_5dc2b_row20_col20\" class=\"data row20 col20\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row20_col21\" class=\"data row20 col21\" >0.04</td>\n",
       "      <td id=\"T_5dc2b_row20_col22\" class=\"data row20 col22\" >-0.38</td>\n",
       "      <td id=\"T_5dc2b_row20_col23\" class=\"data row20 col23\" >-0.48</td>\n",
       "      <td id=\"T_5dc2b_row20_col24\" class=\"data row20 col24\" >-0.41</td>\n",
       "      <td id=\"T_5dc2b_row20_col25\" class=\"data row20 col25\" >0.66</td>\n",
       "      <td id=\"T_5dc2b_row20_col26\" class=\"data row20 col26\" >-0.22</td>\n",
       "      <td id=\"T_5dc2b_row20_col27\" class=\"data row20 col27\" >-0.23</td>\n",
       "      <td id=\"T_5dc2b_row20_col28\" class=\"data row20 col28\" >-0.27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row21\" class=\"row_heading level0 row21\" >MeanNbTradesBySecond</th>\n",
       "      <td id=\"T_5dc2b_row21_col0\" class=\"data row21 col0\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row21_col1\" class=\"data row21 col1\" >0.07</td>\n",
       "      <td id=\"T_5dc2b_row21_col2\" class=\"data row21 col2\" >-0.02</td>\n",
       "      <td id=\"T_5dc2b_row21_col3\" class=\"data row21 col3\" >-0.11</td>\n",
       "      <td id=\"T_5dc2b_row21_col4\" class=\"data row21 col4\" >-0.24</td>\n",
       "      <td id=\"T_5dc2b_row21_col5\" class=\"data row21 col5\" >-0.11</td>\n",
       "      <td id=\"T_5dc2b_row21_col6\" class=\"data row21 col6\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row21_col7\" class=\"data row21 col7\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row21_col8\" class=\"data row21 col8\" >-0.24</td>\n",
       "      <td id=\"T_5dc2b_row21_col9\" class=\"data row21 col9\" >-0.30</td>\n",
       "      <td id=\"T_5dc2b_row21_col10\" class=\"data row21 col10\" >-0.35</td>\n",
       "      <td id=\"T_5dc2b_row21_col11\" class=\"data row21 col11\" >-0.06</td>\n",
       "      <td id=\"T_5dc2b_row21_col12\" class=\"data row21 col12\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row21_col13\" class=\"data row21 col13\" >-0.07</td>\n",
       "      <td id=\"T_5dc2b_row21_col14\" class=\"data row21 col14\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row21_col15\" class=\"data row21 col15\" >-0.09</td>\n",
       "      <td id=\"T_5dc2b_row21_col16\" class=\"data row21 col16\" >-0.22</td>\n",
       "      <td id=\"T_5dc2b_row21_col17\" class=\"data row21 col17\" >-0.27</td>\n",
       "      <td id=\"T_5dc2b_row21_col18\" class=\"data row21 col18\" >-0.36</td>\n",
       "      <td id=\"T_5dc2b_row21_col19\" class=\"data row21 col19\" >-0.16</td>\n",
       "      <td id=\"T_5dc2b_row21_col20\" class=\"data row21 col20\" >0.04</td>\n",
       "      <td id=\"T_5dc2b_row21_col21\" class=\"data row21 col21\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row21_col22\" class=\"data row21 col22\" >-0.12</td>\n",
       "      <td id=\"T_5dc2b_row21_col23\" class=\"data row21 col23\" >-0.11</td>\n",
       "      <td id=\"T_5dc2b_row21_col24\" class=\"data row21 col24\" >-0.16</td>\n",
       "      <td id=\"T_5dc2b_row21_col25\" class=\"data row21 col25\" >-0.21</td>\n",
       "      <td id=\"T_5dc2b_row21_col26\" class=\"data row21 col26\" >0.03</td>\n",
       "      <td id=\"T_5dc2b_row21_col27\" class=\"data row21 col27\" >0.02</td>\n",
       "      <td id=\"T_5dc2b_row21_col28\" class=\"data row21 col28\" >0.03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row22\" class=\"row_heading level0 row22\" >min_dt_TV1</th>\n",
       "      <td id=\"T_5dc2b_row22_col0\" class=\"data row22 col0\" >0.14</td>\n",
       "      <td id=\"T_5dc2b_row22_col1\" class=\"data row22 col1\" >0.06</td>\n",
       "      <td id=\"T_5dc2b_row22_col2\" class=\"data row22 col2\" >-0.20</td>\n",
       "      <td id=\"T_5dc2b_row22_col3\" class=\"data row22 col3\" >0.09</td>\n",
       "      <td id=\"T_5dc2b_row22_col4\" class=\"data row22 col4\" >0.23</td>\n",
       "      <td id=\"T_5dc2b_row22_col5\" class=\"data row22 col5\" >0.09</td>\n",
       "      <td id=\"T_5dc2b_row22_col6\" class=\"data row22 col6\" >0.11</td>\n",
       "      <td id=\"T_5dc2b_row22_col7\" class=\"data row22 col7\" >0.09</td>\n",
       "      <td id=\"T_5dc2b_row22_col8\" class=\"data row22 col8\" >0.24</td>\n",
       "      <td id=\"T_5dc2b_row22_col9\" class=\"data row22 col9\" >0.17</td>\n",
       "      <td id=\"T_5dc2b_row22_col10\" class=\"data row22 col10\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row22_col11\" class=\"data row22 col11\" >0.45</td>\n",
       "      <td id=\"T_5dc2b_row22_col12\" class=\"data row22 col12\" >0.35</td>\n",
       "      <td id=\"T_5dc2b_row22_col13\" class=\"data row22 col13\" >0.44</td>\n",
       "      <td id=\"T_5dc2b_row22_col14\" class=\"data row22 col14\" >0.37</td>\n",
       "      <td id=\"T_5dc2b_row22_col15\" class=\"data row22 col15\" >0.42</td>\n",
       "      <td id=\"T_5dc2b_row22_col16\" class=\"data row22 col16\" >0.28</td>\n",
       "      <td id=\"T_5dc2b_row22_col17\" class=\"data row22 col17\" >0.21</td>\n",
       "      <td id=\"T_5dc2b_row22_col18\" class=\"data row22 col18\" >0.03</td>\n",
       "      <td id=\"T_5dc2b_row22_col19\" class=\"data row22 col19\" >-0.35</td>\n",
       "      <td id=\"T_5dc2b_row22_col20\" class=\"data row22 col20\" >-0.38</td>\n",
       "      <td id=\"T_5dc2b_row22_col21\" class=\"data row22 col21\" >-0.12</td>\n",
       "      <td id=\"T_5dc2b_row22_col22\" class=\"data row22 col22\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row22_col23\" class=\"data row22 col23\" >0.93</td>\n",
       "      <td id=\"T_5dc2b_row22_col24\" class=\"data row22 col24\" >0.99</td>\n",
       "      <td id=\"T_5dc2b_row22_col25\" class=\"data row22 col25\" >-0.35</td>\n",
       "      <td id=\"T_5dc2b_row22_col26\" class=\"data row22 col26\" >0.52</td>\n",
       "      <td id=\"T_5dc2b_row22_col27\" class=\"data row22 col27\" >0.52</td>\n",
       "      <td id=\"T_5dc2b_row22_col28\" class=\"data row22 col28\" >0.51</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row23\" class=\"row_heading level0 row23\" >mean_dt_TV1</th>\n",
       "      <td id=\"T_5dc2b_row23_col0\" class=\"data row23 col0\" >0.09</td>\n",
       "      <td id=\"T_5dc2b_row23_col1\" class=\"data row23 col1\" >0.14</td>\n",
       "      <td id=\"T_5dc2b_row23_col2\" class=\"data row23 col2\" >-0.23</td>\n",
       "      <td id=\"T_5dc2b_row23_col3\" class=\"data row23 col3\" >0.05</td>\n",
       "      <td id=\"T_5dc2b_row23_col4\" class=\"data row23 col4\" >0.31</td>\n",
       "      <td id=\"T_5dc2b_row23_col5\" class=\"data row23 col5\" >0.06</td>\n",
       "      <td id=\"T_5dc2b_row23_col6\" class=\"data row23 col6\" >0.16</td>\n",
       "      <td id=\"T_5dc2b_row23_col7\" class=\"data row23 col7\" >0.08</td>\n",
       "      <td id=\"T_5dc2b_row23_col8\" class=\"data row23 col8\" >0.30</td>\n",
       "      <td id=\"T_5dc2b_row23_col9\" class=\"data row23 col9\" >0.27</td>\n",
       "      <td id=\"T_5dc2b_row23_col10\" class=\"data row23 col10\" >-0.04</td>\n",
       "      <td id=\"T_5dc2b_row23_col11\" class=\"data row23 col11\" >0.45</td>\n",
       "      <td id=\"T_5dc2b_row23_col12\" class=\"data row23 col12\" >0.39</td>\n",
       "      <td id=\"T_5dc2b_row23_col13\" class=\"data row23 col13\" >0.45</td>\n",
       "      <td id=\"T_5dc2b_row23_col14\" class=\"data row23 col14\" >0.40</td>\n",
       "      <td id=\"T_5dc2b_row23_col15\" class=\"data row23 col15\" >0.44</td>\n",
       "      <td id=\"T_5dc2b_row23_col16\" class=\"data row23 col16\" >0.34</td>\n",
       "      <td id=\"T_5dc2b_row23_col17\" class=\"data row23 col17\" >0.27</td>\n",
       "      <td id=\"T_5dc2b_row23_col18\" class=\"data row23 col18\" >0.05</td>\n",
       "      <td id=\"T_5dc2b_row23_col19\" class=\"data row23 col19\" >-0.50</td>\n",
       "      <td id=\"T_5dc2b_row23_col20\" class=\"data row23 col20\" >-0.48</td>\n",
       "      <td id=\"T_5dc2b_row23_col21\" class=\"data row23 col21\" >-0.11</td>\n",
       "      <td id=\"T_5dc2b_row23_col22\" class=\"data row23 col22\" >0.93</td>\n",
       "      <td id=\"T_5dc2b_row23_col23\" class=\"data row23 col23\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row23_col24\" class=\"data row23 col24\" >0.96</td>\n",
       "      <td id=\"T_5dc2b_row23_col25\" class=\"data row23 col25\" >-0.49</td>\n",
       "      <td id=\"T_5dc2b_row23_col26\" class=\"data row23 col26\" >0.40</td>\n",
       "      <td id=\"T_5dc2b_row23_col27\" class=\"data row23 col27\" >0.40</td>\n",
       "      <td id=\"T_5dc2b_row23_col28\" class=\"data row23 col28\" >0.41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row24\" class=\"row_heading level0 row24\" >med_dt_TV1</th>\n",
       "      <td id=\"T_5dc2b_row24_col0\" class=\"data row24 col0\" >0.10</td>\n",
       "      <td id=\"T_5dc2b_row24_col1\" class=\"data row24 col1\" >0.07</td>\n",
       "      <td id=\"T_5dc2b_row24_col2\" class=\"data row24 col2\" >-0.20</td>\n",
       "      <td id=\"T_5dc2b_row24_col3\" class=\"data row24 col3\" >0.08</td>\n",
       "      <td id=\"T_5dc2b_row24_col4\" class=\"data row24 col4\" >0.28</td>\n",
       "      <td id=\"T_5dc2b_row24_col5\" class=\"data row24 col5\" >0.08</td>\n",
       "      <td id=\"T_5dc2b_row24_col6\" class=\"data row24 col6\" >0.13</td>\n",
       "      <td id=\"T_5dc2b_row24_col7\" class=\"data row24 col7\" >0.09</td>\n",
       "      <td id=\"T_5dc2b_row24_col8\" class=\"data row24 col8\" >0.28</td>\n",
       "      <td id=\"T_5dc2b_row24_col9\" class=\"data row24 col9\" >0.24</td>\n",
       "      <td id=\"T_5dc2b_row24_col10\" class=\"data row24 col10\" >-0.04</td>\n",
       "      <td id=\"T_5dc2b_row24_col11\" class=\"data row24 col11\" >0.46</td>\n",
       "      <td id=\"T_5dc2b_row24_col12\" class=\"data row24 col12\" >0.39</td>\n",
       "      <td id=\"T_5dc2b_row24_col13\" class=\"data row24 col13\" >0.46</td>\n",
       "      <td id=\"T_5dc2b_row24_col14\" class=\"data row24 col14\" >0.40</td>\n",
       "      <td id=\"T_5dc2b_row24_col15\" class=\"data row24 col15\" >0.45</td>\n",
       "      <td id=\"T_5dc2b_row24_col16\" class=\"data row24 col16\" >0.33</td>\n",
       "      <td id=\"T_5dc2b_row24_col17\" class=\"data row24 col17\" >0.26</td>\n",
       "      <td id=\"T_5dc2b_row24_col18\" class=\"data row24 col18\" >0.08</td>\n",
       "      <td id=\"T_5dc2b_row24_col19\" class=\"data row24 col19\" >-0.40</td>\n",
       "      <td id=\"T_5dc2b_row24_col20\" class=\"data row24 col20\" >-0.41</td>\n",
       "      <td id=\"T_5dc2b_row24_col21\" class=\"data row24 col21\" >-0.16</td>\n",
       "      <td id=\"T_5dc2b_row24_col22\" class=\"data row24 col22\" >0.99</td>\n",
       "      <td id=\"T_5dc2b_row24_col23\" class=\"data row24 col23\" >0.96</td>\n",
       "      <td id=\"T_5dc2b_row24_col24\" class=\"data row24 col24\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row24_col25\" class=\"data row24 col25\" >-0.39</td>\n",
       "      <td id=\"T_5dc2b_row24_col26\" class=\"data row24 col26\" >0.48</td>\n",
       "      <td id=\"T_5dc2b_row24_col27\" class=\"data row24 col27\" >0.48</td>\n",
       "      <td id=\"T_5dc2b_row24_col28\" class=\"data row24 col28\" >0.48</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row25\" class=\"row_heading level0 row25\" >NbSecondWithAtLeatOneTrade</th>\n",
       "      <td id=\"T_5dc2b_row25_col0\" class=\"data row25 col0\" >0.15</td>\n",
       "      <td id=\"T_5dc2b_row25_col1\" class=\"data row25 col1\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row25_col2\" class=\"data row25 col2\" >0.14</td>\n",
       "      <td id=\"T_5dc2b_row25_col3\" class=\"data row25 col3\" >-0.10</td>\n",
       "      <td id=\"T_5dc2b_row25_col4\" class=\"data row25 col4\" >-0.30</td>\n",
       "      <td id=\"T_5dc2b_row25_col5\" class=\"data row25 col5\" >-0.11</td>\n",
       "      <td id=\"T_5dc2b_row25_col6\" class=\"data row25 col6\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row25_col7\" class=\"data row25 col7\" >-0.13</td>\n",
       "      <td id=\"T_5dc2b_row25_col8\" class=\"data row25 col8\" >-0.22</td>\n",
       "      <td id=\"T_5dc2b_row25_col9\" class=\"data row25 col9\" >-0.24</td>\n",
       "      <td id=\"T_5dc2b_row25_col10\" class=\"data row25 col10\" >0.02</td>\n",
       "      <td id=\"T_5dc2b_row25_col11\" class=\"data row25 col11\" >-0.34</td>\n",
       "      <td id=\"T_5dc2b_row25_col12\" class=\"data row25 col12\" >-0.39</td>\n",
       "      <td id=\"T_5dc2b_row25_col13\" class=\"data row25 col13\" >-0.34</td>\n",
       "      <td id=\"T_5dc2b_row25_col14\" class=\"data row25 col14\" >-0.37</td>\n",
       "      <td id=\"T_5dc2b_row25_col15\" class=\"data row25 col15\" >-0.34</td>\n",
       "      <td id=\"T_5dc2b_row25_col16\" class=\"data row25 col16\" >-0.35</td>\n",
       "      <td id=\"T_5dc2b_row25_col17\" class=\"data row25 col17\" >-0.31</td>\n",
       "      <td id=\"T_5dc2b_row25_col18\" class=\"data row25 col18\" >0.01</td>\n",
       "      <td id=\"T_5dc2b_row25_col19\" class=\"data row25 col19\" >0.63</td>\n",
       "      <td id=\"T_5dc2b_row25_col20\" class=\"data row25 col20\" >0.66</td>\n",
       "      <td id=\"T_5dc2b_row25_col21\" class=\"data row25 col21\" >-0.21</td>\n",
       "      <td id=\"T_5dc2b_row25_col22\" class=\"data row25 col22\" >-0.35</td>\n",
       "      <td id=\"T_5dc2b_row25_col23\" class=\"data row25 col23\" >-0.49</td>\n",
       "      <td id=\"T_5dc2b_row25_col24\" class=\"data row25 col24\" >-0.39</td>\n",
       "      <td id=\"T_5dc2b_row25_col25\" class=\"data row25 col25\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row25_col26\" class=\"data row25 col26\" >-0.24</td>\n",
       "      <td id=\"T_5dc2b_row25_col27\" class=\"data row25 col27\" >-0.26</td>\n",
       "      <td id=\"T_5dc2b_row25_col28\" class=\"data row25 col28\" >-0.32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row26\" class=\"row_heading level0 row26\" >moy_min_dt_TVi_TVj</th>\n",
       "      <td id=\"T_5dc2b_row26_col0\" class=\"data row26 col0\" >-0.01</td>\n",
       "      <td id=\"T_5dc2b_row26_col1\" class=\"data row26 col1\" >-0.02</td>\n",
       "      <td id=\"T_5dc2b_row26_col2\" class=\"data row26 col2\" >-0.19</td>\n",
       "      <td id=\"T_5dc2b_row26_col3\" class=\"data row26 col3\" >0.02</td>\n",
       "      <td id=\"T_5dc2b_row26_col4\" class=\"data row26 col4\" >-0.03</td>\n",
       "      <td id=\"T_5dc2b_row26_col5\" class=\"data row26 col5\" >0.02</td>\n",
       "      <td id=\"T_5dc2b_row26_col6\" class=\"data row26 col6\" >0.03</td>\n",
       "      <td id=\"T_5dc2b_row26_col7\" class=\"data row26 col7\" >0.04</td>\n",
       "      <td id=\"T_5dc2b_row26_col8\" class=\"data row26 col8\" >-0.05</td>\n",
       "      <td id=\"T_5dc2b_row26_col9\" class=\"data row26 col9\" >-0.08</td>\n",
       "      <td id=\"T_5dc2b_row26_col10\" class=\"data row26 col10\" >-0.16</td>\n",
       "      <td id=\"T_5dc2b_row26_col11\" class=\"data row26 col11\" >0.17</td>\n",
       "      <td id=\"T_5dc2b_row26_col12\" class=\"data row26 col12\" >0.15</td>\n",
       "      <td id=\"T_5dc2b_row26_col13\" class=\"data row26 col13\" >0.17</td>\n",
       "      <td id=\"T_5dc2b_row26_col14\" class=\"data row26 col14\" >0.14</td>\n",
       "      <td id=\"T_5dc2b_row26_col15\" class=\"data row26 col15\" >0.16</td>\n",
       "      <td id=\"T_5dc2b_row26_col16\" class=\"data row26 col16\" >0.13</td>\n",
       "      <td id=\"T_5dc2b_row26_col17\" class=\"data row26 col17\" >0.10</td>\n",
       "      <td id=\"T_5dc2b_row26_col18\" class=\"data row26 col18\" >-0.00</td>\n",
       "      <td id=\"T_5dc2b_row26_col19\" class=\"data row26 col19\" >-0.20</td>\n",
       "      <td id=\"T_5dc2b_row26_col20\" class=\"data row26 col20\" >-0.22</td>\n",
       "      <td id=\"T_5dc2b_row26_col21\" class=\"data row26 col21\" >0.03</td>\n",
       "      <td id=\"T_5dc2b_row26_col22\" class=\"data row26 col22\" >0.52</td>\n",
       "      <td id=\"T_5dc2b_row26_col23\" class=\"data row26 col23\" >0.40</td>\n",
       "      <td id=\"T_5dc2b_row26_col24\" class=\"data row26 col24\" >0.48</td>\n",
       "      <td id=\"T_5dc2b_row26_col25\" class=\"data row26 col25\" >-0.24</td>\n",
       "      <td id=\"T_5dc2b_row26_col26\" class=\"data row26 col26\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row26_col27\" class=\"data row26 col27\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row26_col28\" class=\"data row26 col28\" >0.99</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row27\" class=\"row_heading level0 row27\" >moy_med_dt_TVi_TVj</th>\n",
       "      <td id=\"T_5dc2b_row27_col0\" class=\"data row27 col0\" >-0.02</td>\n",
       "      <td id=\"T_5dc2b_row27_col1\" class=\"data row27 col1\" >-0.01</td>\n",
       "      <td id=\"T_5dc2b_row27_col2\" class=\"data row27 col2\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row27_col3\" class=\"data row27 col3\" >0.01</td>\n",
       "      <td id=\"T_5dc2b_row27_col4\" class=\"data row27 col4\" >-0.02</td>\n",
       "      <td id=\"T_5dc2b_row27_col5\" class=\"data row27 col5\" >0.02</td>\n",
       "      <td id=\"T_5dc2b_row27_col6\" class=\"data row27 col6\" >0.04</td>\n",
       "      <td id=\"T_5dc2b_row27_col7\" class=\"data row27 col7\" >0.04</td>\n",
       "      <td id=\"T_5dc2b_row27_col8\" class=\"data row27 col8\" >-0.04</td>\n",
       "      <td id=\"T_5dc2b_row27_col9\" class=\"data row27 col9\" >-0.06</td>\n",
       "      <td id=\"T_5dc2b_row27_col10\" class=\"data row27 col10\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row27_col11\" class=\"data row27 col11\" >0.18</td>\n",
       "      <td id=\"T_5dc2b_row27_col12\" class=\"data row27 col12\" >0.16</td>\n",
       "      <td id=\"T_5dc2b_row27_col13\" class=\"data row27 col13\" >0.17</td>\n",
       "      <td id=\"T_5dc2b_row27_col14\" class=\"data row27 col14\" >0.15</td>\n",
       "      <td id=\"T_5dc2b_row27_col15\" class=\"data row27 col15\" >0.17</td>\n",
       "      <td id=\"T_5dc2b_row27_col16\" class=\"data row27 col16\" >0.14</td>\n",
       "      <td id=\"T_5dc2b_row27_col17\" class=\"data row27 col17\" >0.12</td>\n",
       "      <td id=\"T_5dc2b_row27_col18\" class=\"data row27 col18\" >0.01</td>\n",
       "      <td id=\"T_5dc2b_row27_col19\" class=\"data row27 col19\" >-0.21</td>\n",
       "      <td id=\"T_5dc2b_row27_col20\" class=\"data row27 col20\" >-0.23</td>\n",
       "      <td id=\"T_5dc2b_row27_col21\" class=\"data row27 col21\" >0.02</td>\n",
       "      <td id=\"T_5dc2b_row27_col22\" class=\"data row27 col22\" >0.52</td>\n",
       "      <td id=\"T_5dc2b_row27_col23\" class=\"data row27 col23\" >0.40</td>\n",
       "      <td id=\"T_5dc2b_row27_col24\" class=\"data row27 col24\" >0.48</td>\n",
       "      <td id=\"T_5dc2b_row27_col25\" class=\"data row27 col25\" >-0.26</td>\n",
       "      <td id=\"T_5dc2b_row27_col26\" class=\"data row27 col26\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row27_col27\" class=\"data row27 col27\" >1.00</td>\n",
       "      <td id=\"T_5dc2b_row27_col28\" class=\"data row27 col28\" >0.99</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5dc2b_level0_row28\" class=\"row_heading level0 row28\" >moy_mean_dt_TVi_TVj</th>\n",
       "      <td id=\"T_5dc2b_row28_col0\" class=\"data row28 col0\" >-0.03</td>\n",
       "      <td id=\"T_5dc2b_row28_col1\" class=\"data row28 col1\" >-0.01</td>\n",
       "      <td id=\"T_5dc2b_row28_col2\" class=\"data row28 col2\" >-0.18</td>\n",
       "      <td id=\"T_5dc2b_row28_col3\" class=\"data row28 col3\" >0.01</td>\n",
       "      <td id=\"T_5dc2b_row28_col4\" class=\"data row28 col4\" >0.00</td>\n",
       "      <td id=\"T_5dc2b_row28_col5\" class=\"data row28 col5\" >0.01</td>\n",
       "      <td id=\"T_5dc2b_row28_col6\" class=\"data row28 col6\" >0.05</td>\n",
       "      <td id=\"T_5dc2b_row28_col7\" class=\"data row28 col7\" >0.04</td>\n",
       "      <td id=\"T_5dc2b_row28_col8\" class=\"data row28 col8\" >-0.03</td>\n",
       "      <td id=\"T_5dc2b_row28_col9\" class=\"data row28 col9\" >-0.05</td>\n",
       "      <td id=\"T_5dc2b_row28_col10\" class=\"data row28 col10\" >-0.14</td>\n",
       "      <td id=\"T_5dc2b_row28_col11\" class=\"data row28 col11\" >0.17</td>\n",
       "      <td id=\"T_5dc2b_row28_col12\" class=\"data row28 col12\" >0.16</td>\n",
       "      <td id=\"T_5dc2b_row28_col13\" class=\"data row28 col13\" >0.17</td>\n",
       "      <td id=\"T_5dc2b_row28_col14\" class=\"data row28 col14\" >0.15</td>\n",
       "      <td id=\"T_5dc2b_row28_col15\" class=\"data row28 col15\" >0.16</td>\n",
       "      <td id=\"T_5dc2b_row28_col16\" class=\"data row28 col16\" >0.14</td>\n",
       "      <td id=\"T_5dc2b_row28_col17\" class=\"data row28 col17\" >0.12</td>\n",
       "      <td id=\"T_5dc2b_row28_col18\" class=\"data row28 col18\" >-0.01</td>\n",
       "      <td id=\"T_5dc2b_row28_col19\" class=\"data row28 col19\" >-0.23</td>\n",
       "      <td id=\"T_5dc2b_row28_col20\" class=\"data row28 col20\" >-0.27</td>\n",
       "      <td id=\"T_5dc2b_row28_col21\" class=\"data row28 col21\" >0.03</td>\n",
       "      <td id=\"T_5dc2b_row28_col22\" class=\"data row28 col22\" >0.51</td>\n",
       "      <td id=\"T_5dc2b_row28_col23\" class=\"data row28 col23\" >0.41</td>\n",
       "      <td id=\"T_5dc2b_row28_col24\" class=\"data row28 col24\" >0.48</td>\n",
       "      <td id=\"T_5dc2b_row28_col25\" class=\"data row28 col25\" >-0.32</td>\n",
       "      <td id=\"T_5dc2b_row28_col26\" class=\"data row28 col26\" >0.99</td>\n",
       "      <td id=\"T_5dc2b_row28_col27\" class=\"data row28 col27\" >0.99</td>\n",
       "      <td id=\"T_5dc2b_row28_col28\" class=\"data row28 col28\" >1.00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x1bf23fe0908>"
      ]
     },
     "execution_count": 592,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "corr =x_train.corr()\n",
    "corr.style.background_gradient(cmap='coolwarm').format(precision=2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0279e572",
   "metadata": {},
   "source": [
    "    Plusieurs cases de la correlation map présentent de très fortes corrélations . Les paires fortement corrélées sont les suivantes : "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 593,
   "id": "fcb9ad32",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "10_p_lifetime_cancel  min_lifetime_cancel     0.999806\n",
       "moy_min_dt_TVi_TVj    moy_med_dt_TVi_TVj      0.998630\n",
       "10_p_time_two_events  min_time_two_events     0.998044\n",
       "10_p_lifetime_cancel  25_p_lifetime_cancel    0.995312\n",
       "25_p_lifetime_cancel  min_lifetime_cancel     0.993385\n",
       "moy_mean_dt_TVi_TVj   moy_med_dt_TVi_TVj      0.990539\n",
       "med_lifetime_cancel   mean_lifetime_cancel    0.989939\n",
       "med_dt_TV1            min_dt_TV1              0.986625\n",
       "moy_min_dt_TVi_TVj    moy_mean_dt_TVi_TVj     0.985421\n",
       "75_p_lifetime_cancel  90_p_lifetime_cancel    0.980433\n",
       "mean_lifetime_cancel  75_p_lifetime_cancel    0.977615\n",
       "75_p_time_two_events  mean_time_two_events    0.975361\n",
       "25_p_lifetime_cancel  med_lifetime_cancel     0.969085\n",
       "mean_dt_TV1           med_dt_TV1              0.962421\n",
       "mean_time_two_events  90_p_time_two_events    0.944956\n",
       "75_p_lifetime_cancel  med_lifetime_cancel     0.944306\n",
       "10_p_lifetime_cancel  med_lifetime_cancel     0.942624\n",
       "med_lifetime_cancel   min_lifetime_cancel     0.937073\n",
       "mean_lifetime_cancel  25_p_lifetime_cancel    0.935628\n",
       "25_p_time_two_events  10_p_time_two_events    0.933432\n",
       "min_dt_TV1            mean_dt_TV1             0.928470\n",
       "mean_lifetime_cancel  90_p_lifetime_cancel    0.924705\n",
       "90_p_time_two_events  75_p_time_two_events    0.923141\n",
       "25_p_time_two_events  min_time_two_events     0.909702\n",
       "10_p_lifetime_cancel  mean_lifetime_cancel    0.902921\n",
       "dtype: float64"
      ]
     },
     "execution_count": 593,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "s=x_train.corr().unstack().sort_values(ascending=False).drop_duplicates()\n",
    "s.where(s>0.9).dropna().drop(('OTR','OTR'))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2b500794",
   "metadata": {},
   "source": [
    "####  <font color='red'><u>Commentaire</u></font> :  présence de corrélations allant jusqu'à 99% \n",
    " #### Pour résoudre ce problème de corrélation et sélectionner les variables ,  On va utiliser la régression logistique + Lasso , en optimisant le degré de pénalité par la cross validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 594,
   "id": "f3afca00",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ True,  True,  True, False, False,  True,  True,  True, False,\n",
       "        True,  True,  True, False, False, False, False,  True,  True,\n",
       "       False,  True,  True,  True,  True, False,  True,  True, False,\n",
       "       False,  True])"
      ]
     },
     "execution_count": 594,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# essayer la sélection avec l'hyperparamètre de défaut ( C=1 )\n",
    "sel_ = SelectFromModel(LogisticRegression(C=1, penalty='l1',multi_class='ovr',solver='liblinear'))\n",
    "scaler = StandardScaler()\n",
    "scaler.fit(x_train)  # standardiser la matrice des variables explicatives \n",
    "\n",
    "sel_.fit(scaler.transform(x_train), y_train) # entraîner le modèle pour la sélection des variables\n",
    "sel_.get_support() # afficher la sélection\n",
    "# les valeurs VRAI correspondent aux variables gardées et les variables FAUX correspondent aux variables supprimées "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 595,
   "id": "d535be25",
   "metadata": {},
   "outputs": [],
   "source": [
    "# optimisation de l'hyperparamètre C par la cross validation\n",
    "import numpy as np\n",
    "from sklearn.model_selection import cross_val_score\n",
    "C= [x for x in np.linspace(0.1,5,20)]\n",
    "cv_scores = []\n",
    "for c in C:\n",
    "    logistics_lasso = LogisticRegression(C=c, penalty='l1',multi_class='ovr',solver='liblinear')\n",
    "    scores = cross_val_score(logistics_lasso,scaler.transform(x_train), y_train, cv=3) # cross validation\n",
    "    cv_scores.append(scores.mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 596,
   "id": "6c21770d",
   "metadata": {},
   "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": [
    "plt.plot(C, cv_scores)\n",
    "plt.xlabel('C')\n",
    "plt.ylabel('cross validated score')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 597,
   "id": "779cc6df",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "le degré de pénalisation optimum est:  1.905263157894737 . Il génère un score de  0.8238916256157637\n"
     ]
    }
   ],
   "source": [
    "print('le degré de pénalisation optimum est: ', C[np.array(cv_scores).argmax()] , \n",
    "      '. Il génère un score de ', np.array(cv_scores).max())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 598,
   "id": "626f8e24",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['OTR', 'OCR', 'OMR', 'min_time_two_events', '10_p_time_two_events',\n",
       "       '25_p_time_two_events', '75_p_time_two_events', '90_p_time_two_events',\n",
       "       'max_time_two_events', 'min_lifetime_cancel', '90_p_lifetime_cancel',\n",
       "       'NbTradeVenueMic', 'MaxNbTradesBySecond', 'MeanNbTradesBySecond',\n",
       "       'min_dt_TV1', 'med_dt_TV1', 'NbSecondWithAtLeatOneTrade',\n",
       "       'moy_mean_dt_TVi_TVj'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 598,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Pour l'hyperparamètre optimal = 1.9 , on a la sélection suivante : \n",
    "sel_ = SelectFromModel(LogisticRegression(C=1.9, penalty='l1',multi_class='ovr',solver='liblinear'))\n",
    "sel_.fit(scaler.transform(x_train), y_train) \n",
    "sel_.get_support() # retourne les valeurs VRAI et FAUX \n",
    "x_train.columns[sel_.get_support()] # les variables gardés par le lasso sont les suivants"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 599,
   "id": "7eac95e2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "nbre de variables explicatives total :  29\n",
      "nbre de variables explicatives gardées :  18\n",
      "nbre de variables explicatives supprimées : 11\n"
     ]
    }
   ],
   "source": [
    "print('nbre de variables explicatives total : ', len(x_train.columns))\n",
    "print('nbre de variables explicatives gardées : ',len(x_train.columns[sel_.get_support()]))\n",
    "print ('nbre de variables explicatives supprimées :' , len(x_train.columns)-len(x_train.columns[sel_.get_support()]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 600,
   "id": "5f9c3ce5",
   "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>OTR</th>\n",
       "      <th>OCR</th>\n",
       "      <th>OMR</th>\n",
       "      <th>min_time_two_events</th>\n",
       "      <th>10_p_time_two_events</th>\n",
       "      <th>25_p_time_two_events</th>\n",
       "      <th>75_p_time_two_events</th>\n",
       "      <th>90_p_time_two_events</th>\n",
       "      <th>max_time_two_events</th>\n",
       "      <th>min_lifetime_cancel</th>\n",
       "      <th>90_p_lifetime_cancel</th>\n",
       "      <th>NbTradeVenueMic</th>\n",
       "      <th>MaxNbTradesBySecond</th>\n",
       "      <th>MeanNbTradesBySecond</th>\n",
       "      <th>min_dt_TV1</th>\n",
       "      <th>med_dt_TV1</th>\n",
       "      <th>NbSecondWithAtLeatOneTrade</th>\n",
       "      <th>moy_mean_dt_TVi_TVj</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>7.807554</td>\n",
       "      <td>34.600000</td>\n",
       "      <td>163.730280</td>\n",
       "      <td>8.951034</td>\n",
       "      <td>8.951034</td>\n",
       "      <td>8.951054</td>\n",
       "      <td>685.248281</td>\n",
       "      <td>2623.099829</td>\n",
       "      <td>2629.815306</td>\n",
       "      <td>150.194326</td>\n",
       "      <td>2629.815306</td>\n",
       "      <td>9.200000</td>\n",
       "      <td>17.400000</td>\n",
       "      <td>5.119288</td>\n",
       "      <td>87.691021</td>\n",
       "      <td>130.584267</td>\n",
       "      <td>39.000000</td>\n",
       "      <td>430.459508</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5.291763</td>\n",
       "      <td>26.348230</td>\n",
       "      <td>163.796670</td>\n",
       "      <td>281.143544</td>\n",
       "      <td>285.624589</td>\n",
       "      <td>497.031552</td>\n",
       "      <td>10712.795325</td>\n",
       "      <td>23401.662768</td>\n",
       "      <td>37386.559638</td>\n",
       "      <td>6366.220501</td>\n",
       "      <td>28594.517919</td>\n",
       "      <td>1.948348</td>\n",
       "      <td>14.722523</td>\n",
       "      <td>1.476684</td>\n",
       "      <td>118.461499</td>\n",
       "      <td>400.064941</td>\n",
       "      <td>104.612012</td>\n",
       "      <td>796.776108</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3.666667</td>\n",
       "      <td>9.142857</td>\n",
       "      <td>163.730280</td>\n",
       "      <td>1725.812961</td>\n",
       "      <td>1725.812961</td>\n",
       "      <td>1725.812961</td>\n",
       "      <td>27085.312114</td>\n",
       "      <td>33114.158286</td>\n",
       "      <td>33114.158286</td>\n",
       "      <td>29126.344504</td>\n",
       "      <td>29126.344504</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.571429</td>\n",
       "      <td>2.428571</td>\n",
       "      <td>6455.785804</td>\n",
       "      <td>6578.318506</td>\n",
       "      <td>1.571429</td>\n",
       "      <td>613.857998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3.261453</td>\n",
       "      <td>11.372549</td>\n",
       "      <td>94.474854</td>\n",
       "      <td>341.188804</td>\n",
       "      <td>343.341275</td>\n",
       "      <td>348.091122</td>\n",
       "      <td>6998.170559</td>\n",
       "      <td>13771.469440</td>\n",
       "      <td>14749.875722</td>\n",
       "      <td>4899.629197</td>\n",
       "      <td>7526.619328</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>2.705882</td>\n",
       "      <td>1.899160</td>\n",
       "      <td>3193.790743</td>\n",
       "      <td>4087.296235</td>\n",
       "      <td>3.470588</td>\n",
       "      <td>613.857998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>9.888919</td>\n",
       "      <td>12.624458</td>\n",
       "      <td>163.730280</td>\n",
       "      <td>3669.775350</td>\n",
       "      <td>3669.775354</td>\n",
       "      <td>3705.061185</td>\n",
       "      <td>14333.335571</td>\n",
       "      <td>22125.586960</td>\n",
       "      <td>26159.785097</td>\n",
       "      <td>12149.436822</td>\n",
       "      <td>22112.822545</td>\n",
       "      <td>1.419753</td>\n",
       "      <td>5.769547</td>\n",
       "      <td>1.862654</td>\n",
       "      <td>1173.328002</td>\n",
       "      <td>1545.688415</td>\n",
       "      <td>13.646091</td>\n",
       "      <td>1609.667081</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        OTR        OCR         OMR  min_time_two_events  10_p_time_two_events  \\\n",
       "0  7.807554  34.600000  163.730280             8.951034              8.951034   \n",
       "1  5.291763  26.348230  163.796670           281.143544            285.624589   \n",
       "2  3.666667   9.142857  163.730280          1725.812961           1725.812961   \n",
       "3  3.261453  11.372549   94.474854           341.188804            343.341275   \n",
       "4  9.888919  12.624458  163.730280          3669.775350           3669.775354   \n",
       "\n",
       "   25_p_time_two_events  75_p_time_two_events  90_p_time_two_events  \\\n",
       "0              8.951054            685.248281           2623.099829   \n",
       "1            497.031552          10712.795325          23401.662768   \n",
       "2           1725.812961          27085.312114          33114.158286   \n",
       "3            348.091122           6998.170559          13771.469440   \n",
       "4           3705.061185          14333.335571          22125.586960   \n",
       "\n",
       "   max_time_two_events  min_lifetime_cancel  90_p_lifetime_cancel  \\\n",
       "0          2629.815306           150.194326           2629.815306   \n",
       "1         37386.559638          6366.220501          28594.517919   \n",
       "2         33114.158286         29126.344504          29126.344504   \n",
       "3         14749.875722          4899.629197           7526.619328   \n",
       "4         26159.785097         12149.436822          22112.822545   \n",
       "\n",
       "   NbTradeVenueMic  MaxNbTradesBySecond  MeanNbTradesBySecond   min_dt_TV1  \\\n",
       "0         9.200000            17.400000              5.119288    87.691021   \n",
       "1         1.948348            14.722523              1.476684   118.461499   \n",
       "2         1.000000             3.571429              2.428571  6455.785804   \n",
       "3         1.000000             2.705882              1.899160  3193.790743   \n",
       "4         1.419753             5.769547              1.862654  1173.328002   \n",
       "\n",
       "    med_dt_TV1  NbSecondWithAtLeatOneTrade  moy_mean_dt_TVi_TVj  \n",
       "0   130.584267                   39.000000           430.459508  \n",
       "1   400.064941                  104.612012           796.776108  \n",
       "2  6578.318506                    1.571429           613.857998  \n",
       "3  4087.296235                    3.470588           613.857998  \n",
       "4  1545.688415                   13.646091          1609.667081  "
      ]
     },
     "execution_count": 600,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train_final=x_train[x_train.columns[sel_.get_support()]]\n",
    "x_train_final.head()  # Notre matrice de variables explicatives final "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "50288ef4",
   "metadata": {},
   "source": [
    "<u><font color='red'>Note:</font></u> On n'a pas encore utilisé les données de test et la cross validation pour la sélection de l'hyperparamètre a été faite en splittant le training set . On ne va toucher au testing set qu'à la fin pour évaluer la performance du(des) modèle(s) final(aux)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "456e182a",
   "metadata": {},
   "source": [
    " ### <u>7- Sélection des modèles de classification et optimisation des hyperparamètres "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "533950b0",
   "metadata": {},
   "source": [
    "#### <u>Modèle 1 :SVM</u>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b7667030",
   "metadata": {},
   "source": [
    "Puisqu'il s'agit d'un problème de multi-classification ( 3 classes )  , le SVM standard n'est pas applicable car il est naturellement un classificateur binaire . Pour pouvoir l'appliquer dans notre modèle , on va l'ajuster par la stratégie OVR : One Over The Rest qui remédie au problème de la classification binaire . "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 601,
   "id": "0f9948f5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "score du modèle :  0.88\n"
     ]
    }
   ],
   "source": [
    "# D'abord , les variables non sélectionnées dans le x_train doivent aussi être supprimés du x_test \n",
    "x_test_final=x_test[x_test.columns[sel_.get_support()]]\n",
    "#standardisation des variables explicatives\n",
    "scaler.fit(x_train_final)\n",
    "# on va essayer d'abord avec l'hyperparamètre de défaut C=1\n",
    "model = SVC(kernel='linear', C=1, random_state=42,decision_function_shape='ovr')\n",
    "model.fit(scaler.transform(x_train_final),y_train)\n",
    "y_pred=model.predict(scaler.transform(x_test_final))\n",
    "print('score du modèle : ' , round(accuracy_score(y_test, y_pred, normalize=True),3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 602,
   "id": "ecda5038",
   "metadata": {},
   "outputs": [],
   "source": [
    "# optimisation de l'hyperparamètre\n",
    "C = [x for x in np.linspace(0.01,2,20)]\n",
    "cv_scores=[]\n",
    "for c in C:\n",
    "    model = SVC(kernel='rbf', C=c, random_state=42,decision_function_shape='ovo')\n",
    "    scores = cross_val_score(model,scaler.transform(x_train_final), y_train, cv=5) \n",
    "    cv_scores.append(scores.mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 603,
   "id": "49c48634",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plotter le Rtat de l'optimisation\n",
    "plt.figure(figsize=(6,5))\n",
    "plt.plot(C, cv_scores)\n",
    "plt.xlabel('C')\n",
    "plt.ylabel('cross validated score',fontsize=12)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 604,
   "id": "da610f0e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "le degré de régularisation optimum est:  1.06 . Il génère un score de cross validation de  0.8\n"
     ]
    }
   ],
   "source": [
    "print('le degré de régularisation optimum est: ', round(C[np.array(cv_scores).argmax()],2) , \n",
    "      '. Il génère un score de cross validation de ', round(np.array(cv_scores).max(),2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 605,
   "id": "e221b653",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "l'accuracy du modèle est :  0.892\n"
     ]
    }
   ],
   "source": [
    "# Avec l'hyperparamètre sélectionné , évaluons le modèle maintenant sur le testing set \n",
    "SVM = SVC(kernel='rbf', C=1.06, random_state=42,decision_function_shape='ovo')\n",
    "SVM.fit(scaler.transform(x_train_final),y_train)\n",
    "y_pred=SVM.predict(scaler.transform(x_test_final))\n",
    "print ( \"l'accuracy du modèle est : \" , round(accuracy_score(y_test, y_pred, normalize=True),3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 606,
   "id": "19719b05",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Pour améliorer le modèle , essayons avec les différents kernels \n",
    "# Pour cela , on va optimiser le couple de paramètres ( kernel , C) par la cross validation \n",
    "C = [x for x in np.linspace(0.01,1,20)]\n",
    "kernel=['linear','poly','sigmoid','rbf']\n",
    "cv_scores={}\n",
    "for c in C:\n",
    "    for k in kernel:\n",
    "        model = SVC(kernel=k, C=c, random_state=42,decision_function_shape='ovo')\n",
    "        scores = cross_val_score(model,scaler.transform(x_train_final), y_train, cv=5) \n",
    "        cv_scores[(k,c)]=scores.mean()\n",
    "#cv_scores est un dictionnaire de la forme suivante : { ('linear',0.01) : 0.67 , ('poly',0.01): 0.82 ....... } avec les clés \n",
    "# étant des tuples formés par les paramètres à optimiser et les valeurs étant les scores cross validés correspondants ."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 607,
   "id": "d84c16ea",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "le couple qui maximise le score cross validé est : ('linear', 0.21842105263157896)\n"
     ]
    }
   ],
   "source": [
    "print('le couple qui maximise le score cross validé est :' , max(cv_scores,key=cv_scores.get) )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 608,
   "id": "b9072416",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "score du modèle :  0.867\n"
     ]
    }
   ],
   "source": [
    "model = SVC(kernel='linear', C=0.21, random_state=42,decision_function_shape='ovr')\n",
    "model.fit(scaler.transform(x_train_final),y_train)\n",
    "y_pred=model.predict(scaler.transform(x_test_final))\n",
    "print('score du modèle : ' , round(accuracy_score(y_test, y_pred, normalize=True),3) )"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "58559ac5",
   "metadata": {},
   "source": [
    " Pour le meilleur couple (kernel , c) = ('linear' , 0.21 ) , l'accuracy finale du SVM est 86.7% . "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 609,
   "id": "0727c4fa",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Evaluons le modèle par d'autres metrics : \n",
    "def conf_matrix_class_report(y_test,y_pred,rtat_évaluation):\n",
    "    y_test_decoded=y_test.replace({0:'HFT', 1:'NON HFT', 2: 'MIX'})\n",
    "    y_pred_decoded=np.array(pd.Series(y_pred).replace({0:'HFT', 1:'NON HFT', 2: 'MIX'}))\n",
    "    if rtat_évaluation=='confusion_matrix' :\n",
    "        rtat=pd.crosstab(y_test_decoded,y_pred_decoded,rownames=['Actual'],colnames=['Predicted'])\n",
    "    elif rtat_évaluation=='classification_report':\n",
    "        rtat=classification_report(y_test_decoded,y_pred_decoded)\n",
    "    print(rtat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 610,
   "id": "3a79e810",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted  HFT  MIX  NON HFT\n",
      "Actual                      \n",
      "HFT          9    2        4\n",
      "MIX          1   20        3\n",
      "NON HFT      0    1       43\n"
     ]
    }
   ],
   "source": [
    "conf_matrix_class_report(y_test,y_pred,'confusion_matrix')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 611,
   "id": "220ec8ed",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Détail du score : 43 NON HFT correctement classifiés , 19 MIX correctement classifiés , 8 HFT correctement classifié soit un\n",
    "# total de 72 observations correctes parmi 83 . "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 612,
   "id": "8d9ab08d",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "         HFT       0.90      0.60      0.72        15\n",
      "         MIX       0.87      0.83      0.85        24\n",
      "     NON HFT       0.86      0.98      0.91        44\n",
      "\n",
      "    accuracy                           0.87        83\n",
      "   macro avg       0.88      0.80      0.83        83\n",
      "weighted avg       0.87      0.87      0.86        83\n",
      "\n"
     ]
    }
   ],
   "source": [
    "conf_matrix_class_report(y_test,y_pred,'classification_report')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5de652d3",
   "metadata": {},
   "source": [
    " <u>**<font color='red'>Cross validation pour un score moyen plus représentatif</font>**</u>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 613,
   "id": "4dfc2a81",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7607843137254903"
      ]
     },
     "execution_count": 613,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# on effectue la cross validation sur tout le dataset \n",
    "x=pd.concat([x_train,x_test],axis=0)\n",
    "y=pd.concat([y_train,y_test],axis=0)\n",
    "scores = cross_val_score(model,x, y, cv=5) \n",
    "scores.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "da4d75d6",
   "metadata": {},
   "source": [
    "#### <u>Modèle 2: KNN </u>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 614,
   "id": "f9754828",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "score du modèle est :  0.783\n"
     ]
    }
   ],
   "source": [
    "knn = KNeighborsClassifier(n_neighbors=6)\n",
    "scaler.fit(x_train_final) # standardisation des variables explicatives\n",
    "knn.fit(scaler.transform(x_train_final),y_train)\n",
    "y_pred=knn.predict(scaler.transform(x_test_final))\n",
    "print('score du modèle est : ' ,round(accuracy_score(y_test, y_pred, normalize=True),3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 615,
   "id": "d9ed02b0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# optimisation de l'hyperparamètre n_neighbors par la cross validation\n",
    "neighbors = [x for x in range(1,50)]\n",
    "cv_scores = []\n",
    "for k in neighbors:\n",
    "    knn = KNeighborsClassifier(n_neighbors=k)\n",
    "    scores = cross_val_score(knn,scaler.transform(x_train_final), y_train, cv=5) \n",
    "    cv_scores.append(scores.mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 616,
   "id": "2ae93356",
   "metadata": {},
   "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": [
    "plt.plot(neighbors, cv_scores)\n",
    "plt.xlabel('Number of Neighbors K')\n",
    "plt.ylabel('cross validated score')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 617,
   "id": "b7d97cc1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "le nbre de voisins optimal est :  3 . Il génère un score de cross validation de  0.7647058823529412\n"
     ]
    }
   ],
   "source": [
    "print('le nbre de voisins optimal est : ', neighbors[np.array(cv_scores).argmax()] , \n",
    "      '. Il génère un score de cross validation de ', np.array(cv_scores).max())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 618,
   "id": "74422809",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8554216867469879"
      ]
     },
     "execution_count": 618,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "knn = KNeighborsClassifier(n_neighbors=3)\n",
    "knn.fit(scaler.transform(x_train_final),y_train)\n",
    "y_pred=knn.predict(scaler.transform(x_test_final))\n",
    "accuracy_score(y_test, y_pred, normalize=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 619,
   "id": "f2435565",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted  HFT  MIX  NON HFT\n",
      "Actual                      \n",
      "HFT         11    2        2\n",
      "MIX          3   17        4\n",
      "NON HFT      1    0       43\n"
     ]
    }
   ],
   "source": [
    "conf_matrix_class_report(y_test,y_pred,'confusion_matrix')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 620,
   "id": "72ad2800",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 43 NON HFT correctement classifiés , 17 MIX correctement classifiés , 11 HFT correctement classifiés soit un\n",
    "# total de 71 observations correctes parmi 83. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 621,
   "id": "895603d6",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "         HFT       0.73      0.73      0.73        15\n",
      "         MIX       0.89      0.71      0.79        24\n",
      "     NON HFT       0.88      0.98      0.92        44\n",
      "\n",
      "    accuracy                           0.86        83\n",
      "   macro avg       0.84      0.81      0.82        83\n",
      "weighted avg       0.86      0.86      0.85        83\n",
      "\n"
     ]
    }
   ],
   "source": [
    "conf_matrix_class_report(y_test,y_pred,'classification_report')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fbf8f3e9",
   "metadata": {},
   "source": [
    " <u>**<font color='red'>Cross validation pour un score moyen plus représentatif</font>**</u>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 622,
   "id": "30c4b6a9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7679144385026737"
      ]
     },
     "execution_count": 622,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scores = cross_val_score(knn,x, y, cv=5) \n",
    "scores.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3f259b75",
   "metadata": {},
   "source": [
    "#### <u>Modèle 3: Linear Discriminant analysis </u>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 623,
   "id": "d8b18ac2",
   "metadata": {},
   "outputs": [],
   "source": [
    "# choix du paramètre solver\n",
    "sol= ['svd', 'lsqr', 'eigen']\n",
    "cv_scores = []\n",
    "for i in sol:\n",
    "    LDA = LinearDiscriminantAnalysis(solver=i)\n",
    "    scores = cross_val_score(LDA,x_train_final, y_train, cv=5) \n",
    "    cv_scores.append(scores.mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 624,
   "id": "293ea732",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0.7176470588235293, 0.7176470588235293, 0.7176470588235293]"
      ]
     },
     "execution_count": 624,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cv_scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 625,
   "id": "bec9b56e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "score du modèle est :  0.843\n"
     ]
    }
   ],
   "source": [
    "# Evaluer la performance du modèle sur le testing set \n",
    "LDA = LinearDiscriminantAnalysis(solver='eigen')\n",
    "LDA.fit(x_train_final,y_train)\n",
    "y_pred=LDA.predict(x_test_final)\n",
    "print('score du modèle est : ' , round(accuracy_score(y_test, y_pred, normalize=True),3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 626,
   "id": "e991c74b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted  HFT  MIX  NON HFT\n",
      "Actual                      \n",
      "HFT          8    2        5\n",
      "MIX          1   20        3\n",
      "NON HFT      1    1       42\n"
     ]
    }
   ],
   "source": [
    "conf_matrix_class_report(y_test,y_pred,'confusion_matrix')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 627,
   "id": "31c32615",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "         HFT       0.80      0.53      0.64        15\n",
      "         MIX       0.87      0.83      0.85        24\n",
      "     NON HFT       0.84      0.95      0.89        44\n",
      "\n",
      "    accuracy                           0.84        83\n",
      "   macro avg       0.84      0.77      0.79        83\n",
      "weighted avg       0.84      0.84      0.84        83\n",
      "\n"
     ]
    }
   ],
   "source": [
    "conf_matrix_class_report(y_test,y_pred,'classification_report')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d5a4f098",
   "metadata": {},
   "source": [
    " <u>**<font color='red'>Cross validation pour un score moyen plus représentatif</font>**</u>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 628,
   "id": "875f0a26",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8333333333333334"
      ]
     },
     "execution_count": 628,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scores = cross_val_score(LDA,x, y, cv=5) \n",
    "scores.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "137d3fe7",
   "metadata": {},
   "source": [
    "#### <u>Modèle 4: Naive Bayes </u>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 629,
   "id": "1efca407",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "score du modèle sur le training set est :  0.706\n",
      "score du modèle sur le testing set est :  0.651\n"
     ]
    }
   ],
   "source": [
    "# caractéristiques du modèle : features indépendants / marche mieux que les autres classifiers pour des datasets de taille \n",
    "# réduite ( cas de notre dataset qui s'est considérablemnt réduit après le regroupement des traders ) \n",
    "gnb = GaussianNB()\n",
    "gnb.fit(x_train_final, y_train)\n",
    "y_pred=gnb.predict(x_test_final)\n",
    "y_pred_train=gnb.predict(x_train_final)\n",
    "print('score du modèle sur le training set est : ' , round(accuracy_score(y_train, y_pred_train, normalize=True),3))\n",
    "print('score du modèle sur le testing set est : ' , round(accuracy_score(y_test, y_pred, normalize=True),3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 630,
   "id": "8f49baa1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "         HFT       0.52      0.73      0.61        15\n",
      "         MIX       0.56      0.79      0.66        24\n",
      "     NON HFT       0.86      0.55      0.67        44\n",
      "\n",
      "    accuracy                           0.65        83\n",
      "   macro avg       0.65      0.69      0.64        83\n",
      "weighted avg       0.71      0.65      0.65        83\n",
      "\n"
     ]
    }
   ],
   "source": [
    "conf_matrix_class_report(y_test,y_pred,'classification_report')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7529d5e6",
   "metadata": {},
   "source": [
    " <u>**<font color='red'>Cross validation pour un score moyen plus représentatif</font>**</u>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 631,
   "id": "ae668ff5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.5944741532976827"
      ]
     },
     "execution_count": 631,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scores = cross_val_score(gnb,x, y, cv=5) \n",
    "scores.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b4a01a0b",
   "metadata": {},
   "source": [
    " ### <u>8- Prédiction sur le testing set donné de l'AMF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 632,
   "id": "d469fe46",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\ASUS\\anaconda3\\lib\\site-packages\\ipykernel_launcher.py:13: 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: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  del sys.path[0]\n"
     ]
    }
   ],
   "source": [
    "test=pd.read_csv('AMF_test_X.csv')\n",
    "test=test.drop(['Index','Share','Day'],axis=1)\n",
    "test=moy_tvi_tvj(test)\n",
    "test[['OMR','OTR']]=test[['OMR','OTR']].fillna(moy_stockés1)\n",
    "test[lst_names]=test[lst_names].fillna(moy_stockés2)\n",
    "test=test.groupby(['Trader']).mean()\n",
    "test=test[test.columns[sel_.get_support()]]\n",
    "y_pred=LDA.predict(test)\n",
    "y_pred_df=pd.DataFrame(y_pred)\n",
    "y_pred_df.head()\n",
    "test1=test.reset_index()\n",
    "trader=test1[['Trader']]\n",
    "trader['Type']=y_pred_df\n",
    "trader1=trader.replace({'Type':{0:'HFT',1:'NON HFT',2:'MIX'}})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 633,
   "id": "e2259d9b",
   "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>Trader</th>\n",
       "      <th>Type</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Adelaide</td>\n",
       "      <td>NON HFT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Alana</td>\n",
       "      <td>NON HFT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Alcmene</td>\n",
       "      <td>NON HFT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Alice</td>\n",
       "      <td>HFT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Alices Sister</td>\n",
       "      <td>NON HFT</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          Trader     Type\n",
       "0       Adelaide  NON HFT\n",
       "1          Alana  NON HFT\n",
       "2        Alcmene  NON HFT\n",
       "3          Alice      HFT\n",
       "4  Alices Sister  NON HFT"
      ]
     },
     "execution_count": 633,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trader1.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 634,
   "id": "57ff2a50",
   "metadata": {},
   "outputs": [],
   "source": [
    "trader1.to_csv('résultat3.csv',index=False)"
   ]
  }
 ],
 "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": "3.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}