Home_Credit_MS.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# PREVENDO O RISCO DE INADIMPLÊNCIA NO CRÉDITO RESIDENCIAL\n",
    "\n",
    "### MODEL SELECTION AND TUNNING\n",
    "\n",
    "#### Udacity - Engenheiro de Machine Learning\n",
    "Vinicius Ferreira Santos - vyniciuss@gmail.com\n",
    "\n",
    "## HOME CREDIT\n",
    "\n",
    "\n",
    "### Importando as bibliotecas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vynic\\Anaconda3\\envs\\py36\\lib\\site-packages\\sklearn\\cross_validation.py:41: DeprecationWarning:\n",
      "\n",
      "This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. Also note that the interface of the new CV iterators are different from that of this module. This module will be removed in 0.20.\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<script>requirejs.config({paths: { 'plotly': ['https://cdn.plot.ly/plotly-latest.min']},});if(!window.Plotly) {{require(['plotly'],function(plotly) {window.Plotly=plotly;});}}</script>"
      ],
      "text/vnd.plotly.v1+html": [
       "<script>requirejs.config({paths: { 'plotly': ['https://cdn.plot.ly/plotly-latest.min']},});if(!window.Plotly) {{require(['plotly'],function(plotly) {window.Plotly=plotly;});}}</script>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<script>requirejs.config({paths: { 'plotly': ['https://cdn.plot.ly/plotly-latest.min']},});if(!window.Plotly) {{require(['plotly'],function(plotly) {window.Plotly=plotly;});}}</script>"
      ],
      "text/vnd.plotly.v1+html": [
       "<script>requirejs.config({paths: { 'plotly': ['https://cdn.plot.ly/plotly-latest.min']},});if(!window.Plotly) {{require(['plotly'],function(plotly) {window.Plotly=plotly;});}}</script>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vynic\\Anaconda3\\envs\\py36\\lib\\site-packages\\sklearn\\ensemble\\weight_boosting.py:29: DeprecationWarning:\n",
      "\n",
      "numpy.core.umath_tests is an internal NumPy module and should not be imported. It will be removed in a future NumPy release.\n",
      "\n",
      "C:\\Users\\vynic\\Anaconda3\\envs\\py36\\lib\\site-packages\\sklearn\\grid_search.py:42: DeprecationWarning:\n",
      "\n",
      "This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. This module will be removed in 0.20.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from matplotlib.legend_handler import HandlerLine2D\n",
    "from helper import *\n",
    "from plotly.offline import init_notebook_mode, iplot, plot\n",
    "import plotly.graph_objs as go\n",
    "import plotly.plotly as py\n",
    "from plotly import tools\n",
    "init_notebook_mode(connected=True)\n",
    "# ================================================== #\n",
    "#               Classifiers                          #\n",
    "# ================================================== #\n",
    "from sklearn.ensemble import GradientBoostingClassifier, AdaBoostClassifier, RandomForestClassifier\n",
    "# ================================================== #\n",
    "#               Validation and Test                  #\n",
    "# ================================================== #\n",
    "from sklearn.metrics import roc_curve, roc_auc_score, accuracy_score, confusion_matrix\n",
    "from sklearn import cross_validation\n",
    "from sklearn.model_selection import train_test_split\n",
    "# ===================================================#\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.grid_search import GridSearchCV\n",
    "from time import time\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "path = \"plots/\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## CARREGANDO OS DADOS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>Unnamed: 0</th>\n",
       "      <th>CNT_CHILDREN</th>\n",
       "      <th>AMT_INCOME_TOTAL</th>\n",
       "      <th>AMT_CREDIT</th>\n",
       "      <th>AMT_ANNUITY</th>\n",
       "      <th>AMT_GOODS_PRICE</th>\n",
       "      <th>REGION_POPULATION_RELATIVE</th>\n",
       "      <th>DAYS_BIRTH</th>\n",
       "      <th>DAYS_EMPLOYED</th>\n",
       "      <th>DAYS_REGISTRATION</th>\n",
       "      <th>...</th>\n",
       "      <th>ORGANIZATION_TYPE_Transport: type 2</th>\n",
       "      <th>ORGANIZATION_TYPE_Transport: type 3</th>\n",
       "      <th>ORGANIZATION_TYPE_Transport: type 4</th>\n",
       "      <th>ORGANIZATION_TYPE_University</th>\n",
       "      <th>ORGANIZATION_TYPE_XNA</th>\n",
       "      <th>EMERGENCYSTATE_MODE_No</th>\n",
       "      <th>EMERGENCYSTATE_MODE_Yes</th>\n",
       "      <th>EMERGENCYSTATE_MODE_nan</th>\n",
       "      <th>SK_ID_CURR</th>\n",
       "      <th>TARGET</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>-0.583687</td>\n",
       "      <td>0.783762</td>\n",
       "      <td>-0.464117</td>\n",
       "      <td>-0.098525</td>\n",
       "      <td>-0.47329</td>\n",
       "      <td>-0.055568</td>\n",
       "      <td>1.50688</td>\n",
       "      <td>0.894907</td>\n",
       "      <td>0.377004</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>100002</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1 rows × 260 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0  CNT_CHILDREN  AMT_INCOME_TOTAL  AMT_CREDIT  AMT_ANNUITY  \\\n",
       "0           0     -0.583687          0.783762   -0.464117    -0.098525   \n",
       "\n",
       "   AMT_GOODS_PRICE  REGION_POPULATION_RELATIVE  DAYS_BIRTH  DAYS_EMPLOYED  \\\n",
       "0         -0.47329                   -0.055568     1.50688       0.894907   \n",
       "\n",
       "   DAYS_REGISTRATION   ...    ORGANIZATION_TYPE_Transport: type 2  \\\n",
       "0           0.377004   ...                                    0.0   \n",
       "\n",
       "   ORGANIZATION_TYPE_Transport: type 3  ORGANIZATION_TYPE_Transport: type 4  \\\n",
       "0                                  0.0                                  0.0   \n",
       "\n",
       "   ORGANIZATION_TYPE_University  ORGANIZATION_TYPE_XNA  \\\n",
       "0                           0.0                    0.0   \n",
       "\n",
       "   EMERGENCYSTATE_MODE_No  EMERGENCYSTATE_MODE_Yes  EMERGENCYSTATE_MODE_nan  \\\n",
       "0                     1.0                      0.0                      0.0   \n",
       "\n",
       "   SK_ID_CURR  TARGET  \n",
       "0      100002       1  \n",
       "\n",
       "[1 rows x 260 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#carregando os dados de treinamento\n",
    "data_train = pd.read_csv(\"data/application_train_new.csv\")\n",
    "\n",
    "data_test = pd.read_csv(\"data/application_test_new.csv\")\n",
    "display(data_train.head(n=1))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Tunning** é o processo de buscar os melhores hiperparâmetros para um determinado estimador. Para facilitar o trabalho, tal processo será executado com o auxílio do GridSearchCV que, como descrito na documentação do sklearn, realiza uma busca exaustiva sobre valores de parâmetros especificados para um estimador. Como podemos perceber, esse é um processo caro e lento de se realizar com computadores pessoais se a massa de dados for muito grande. Dessa forma, irei separar uma amostra de 16 mil linhas, onde 10 mil será para treino e 6 mil para teste, afim de facilitar o processo.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>CNT_CHILDREN</th>\n",
       "      <th>AMT_INCOME_TOTAL</th>\n",
       "      <th>AMT_CREDIT</th>\n",
       "      <th>AMT_ANNUITY</th>\n",
       "      <th>AMT_GOODS_PRICE</th>\n",
       "      <th>REGION_POPULATION_RELATIVE</th>\n",
       "      <th>DAYS_BIRTH</th>\n",
       "      <th>DAYS_EMPLOYED</th>\n",
       "      <th>DAYS_REGISTRATION</th>\n",
       "      <th>DAYS_ID_PUBLISH</th>\n",
       "      <th>...</th>\n",
       "      <th>ORGANIZATION_TYPE_Trade: type 7</th>\n",
       "      <th>ORGANIZATION_TYPE_Transport: type 1</th>\n",
       "      <th>ORGANIZATION_TYPE_Transport: type 2</th>\n",
       "      <th>ORGANIZATION_TYPE_Transport: type 3</th>\n",
       "      <th>ORGANIZATION_TYPE_Transport: type 4</th>\n",
       "      <th>ORGANIZATION_TYPE_University</th>\n",
       "      <th>ORGANIZATION_TYPE_XNA</th>\n",
       "      <th>EMERGENCYSTATE_MODE_No</th>\n",
       "      <th>EMERGENCYSTATE_MODE_Yes</th>\n",
       "      <th>EMERGENCYSTATE_MODE_nan</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>245895</th>\n",
       "      <td>2.548310</td>\n",
       "      <td>0.855572</td>\n",
       "      <td>-0.295497</td>\n",
       "      <td>2.214159</td>\n",
       "      <td>-0.235201</td>\n",
       "      <td>-0.905926</td>\n",
       "      <td>0.627866</td>\n",
       "      <td>0.800316</td>\n",
       "      <td>1.242524</td>\n",
       "      <td>-0.869720</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>98194</th>\n",
       "      <td>-0.583687</td>\n",
       "      <td>1.501855</td>\n",
       "      <td>2.042883</td>\n",
       "      <td>1.908359</td>\n",
       "      <td>2.447259</td>\n",
       "      <td>-1.163509</td>\n",
       "      <td>0.288497</td>\n",
       "      <td>0.513517</td>\n",
       "      <td>0.962833</td>\n",
       "      <td>-1.028056</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36463</th>\n",
       "      <td>-0.583687</td>\n",
       "      <td>0.783762</td>\n",
       "      <td>-0.210865</td>\n",
       "      <td>1.094133</td>\n",
       "      <td>0.034632</td>\n",
       "      <td>1.519881</td>\n",
       "      <td>-0.428509</td>\n",
       "      <td>0.893393</td>\n",
       "      <td>0.704988</td>\n",
       "      <td>1.015737</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>249923</th>\n",
       "      <td>-0.583687</td>\n",
       "      <td>1.501855</td>\n",
       "      <td>-0.899268</td>\n",
       "      <td>-0.078784</td>\n",
       "      <td>-0.917721</td>\n",
       "      <td>2.486791</td>\n",
       "      <td>-0.822415</td>\n",
       "      <td>0.233528</td>\n",
       "      <td>-1.784354</td>\n",
       "      <td>-0.108515</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>158389</th>\n",
       "      <td>-0.583687</td>\n",
       "      <td>-0.652422</td>\n",
       "      <td>-0.746195</td>\n",
       "      <td>-0.830062</td>\n",
       "      <td>-0.885976</td>\n",
       "      <td>-0.051024</td>\n",
       "      <td>-0.983048</td>\n",
       "      <td>0.540759</td>\n",
       "      <td>-0.672487</td>\n",
       "      <td>1.656369</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 257 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        CNT_CHILDREN  AMT_INCOME_TOTAL  AMT_CREDIT  AMT_ANNUITY  \\\n",
       "245895      2.548310          0.855572   -0.295497     2.214159   \n",
       "98194      -0.583687          1.501855    2.042883     1.908359   \n",
       "36463      -0.583687          0.783762   -0.210865     1.094133   \n",
       "249923     -0.583687          1.501855   -0.899268    -0.078784   \n",
       "158389     -0.583687         -0.652422   -0.746195    -0.830062   \n",
       "\n",
       "        AMT_GOODS_PRICE  REGION_POPULATION_RELATIVE  DAYS_BIRTH  \\\n",
       "245895        -0.235201                   -0.905926    0.627866   \n",
       "98194          2.447259                   -1.163509    0.288497   \n",
       "36463          0.034632                    1.519881   -0.428509   \n",
       "249923        -0.917721                    2.486791   -0.822415   \n",
       "158389        -0.885976                   -0.051024   -0.983048   \n",
       "\n",
       "        DAYS_EMPLOYED  DAYS_REGISTRATION  DAYS_ID_PUBLISH  \\\n",
       "245895       0.800316           1.242524        -0.869720   \n",
       "98194        0.513517           0.962833        -1.028056   \n",
       "36463        0.893393           0.704988         1.015737   \n",
       "249923       0.233528          -1.784354        -0.108515   \n",
       "158389       0.540759          -0.672487         1.656369   \n",
       "\n",
       "                 ...             ORGANIZATION_TYPE_Trade: type 7  \\\n",
       "245895           ...                                         0.0   \n",
       "98194            ...                                         0.0   \n",
       "36463            ...                                         0.0   \n",
       "249923           ...                                         0.0   \n",
       "158389           ...                                         0.0   \n",
       "\n",
       "        ORGANIZATION_TYPE_Transport: type 1  \\\n",
       "245895                                  0.0   \n",
       "98194                                   0.0   \n",
       "36463                                   0.0   \n",
       "249923                                  0.0   \n",
       "158389                                  0.0   \n",
       "\n",
       "        ORGANIZATION_TYPE_Transport: type 2  \\\n",
       "245895                                  0.0   \n",
       "98194                                   0.0   \n",
       "36463                                   0.0   \n",
       "249923                                  0.0   \n",
       "158389                                  0.0   \n",
       "\n",
       "        ORGANIZATION_TYPE_Transport: type 3  \\\n",
       "245895                                  0.0   \n",
       "98194                                   0.0   \n",
       "36463                                   0.0   \n",
       "249923                                  0.0   \n",
       "158389                                  0.0   \n",
       "\n",
       "        ORGANIZATION_TYPE_Transport: type 4  ORGANIZATION_TYPE_University  \\\n",
       "245895                                  0.0                           0.0   \n",
       "98194                                   0.0                           0.0   \n",
       "36463                                   0.0                           0.0   \n",
       "249923                                  0.0                           0.0   \n",
       "158389                                  0.0                           0.0   \n",
       "\n",
       "        ORGANIZATION_TYPE_XNA  EMERGENCYSTATE_MODE_No  \\\n",
       "245895                    0.0                     0.0   \n",
       "98194                     0.0                     1.0   \n",
       "36463                     0.0                     0.0   \n",
       "249923                    0.0                     1.0   \n",
       "158389                    0.0                     1.0   \n",
       "\n",
       "        EMERGENCYSTATE_MODE_Yes  EMERGENCYSTATE_MODE_nan  \n",
       "245895                      0.0                      1.0  \n",
       "98194                       0.0                      0.0  \n",
       "36463                       0.0                      1.0  \n",
       "249923                      0.0                      0.0  \n",
       "158389                      0.0                      0.0  \n",
       "\n",
       "[5 rows x 257 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features = data_train.sample(n = 16000, random_state = 42)\n",
    "labels = features['TARGET']\n",
    "features = features.drop(columns = ['TARGET', 'SK_ID_CURR', 'Unnamed: 0'])\n",
    "features_test = data_test.drop(columns = ['SK_ID_CURR', 'Unnamed: 0'])\n",
    "features.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Separando em dados treino e teste"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train, x_test, y_train, y_test = train_test_split(features, labels,\n",
    "                                                    test_size=6000,\n",
    "                                                    random_state=42,\n",
    "                                                    stratify=labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Escolha e Afinamento de um Modelo\n",
    "\n",
    "### Treinar uma variedade de classificadores\n",
    "\n",
    "Como estou com um problema de natureza superviosionada e com classes desbalanceadas, usarei os seguintes algorítmos:\n",
    "\n",
    "- Adaboost\n",
    "- Gradient boosting\n",
    "- RandomForestClassifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### RandomForestClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tempo de execução RandomForestClassifier: 0.299 s\n",
      "Train Accuracy: 0.9854%\n",
      "Train Score Roc: 99.992%\n",
      "Train CV Score : Mean - 0.6109644 | Std - 0.02210798 | Min - 0.5724992 | Max - 0.6321221\n",
      "------------------------------------------------------------------------------------------------\n",
      "Test Accuracy: 0.918%\n",
      "Test Score Roc: 60.291%\n",
      "Test CV Score : Mean - 0.5748825 | Std - 0.03729735 | Min - 0.5378929 | Max - 0.6393709\n"
     ]
    },
    {
     "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": [
    "classifier = RandomForestClassifier()\n",
    "t0 = time()\n",
    "classifier.fit(x_train, y_train)\n",
    "print(\"Tempo de execução RandomForestClassifier:\", round(time()-t0, 3), \"s\")\n",
    "draw_roc_score(classifier, x_train, y_train, x_test, y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Adaboost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tempo de execução Adaboost: 3.083 s\n",
      "Train Accuracy: 0.9194%\n",
      "Train Score Roc: 79.307%\n",
      "Train CV Score : Mean - 0.717786 | Std - 0.01726962 | Min - 0.6923086 | Max - 0.7465727\n",
      "------------------------------------------------------------------------------------------------\n",
      "Test Accuracy: 0.9155%\n",
      "Test Score Roc: 71.711%\n",
      "Test CV Score : Mean - 0.6796754 | Std - 0.02830649 | Min - 0.6277808 | Max - 0.7087716\n"
     ]
    },
    {
     "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": [
    "classifier = AdaBoostClassifier()\n",
    "t0 = time()\n",
    "classifier.fit(x_train, y_train)\n",
    "print(\"Tempo de execução Adaboost:\", round(time()-t0, 3), \"s\")\n",
    "draw_roc_score(classifier, x_train, y_train, x_test, y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Gradient Boosting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tempo de execução Gradient Boosting: 7.672 s\n",
      "Train Accuracy: 0.9245%\n",
      "Train Score Roc: 85.584%\n",
      "Train CV Score : Mean - 0.7282286 | Std - 0.008040625 | Min - 0.7189369 | Max - 0.7417468\n",
      "------------------------------------------------------------------------------------------------\n",
      "Test Accuracy: 0.9161666666666667%\n",
      "Test Score Roc: 73.12%\n",
      "Test CV Score : Mean - 0.7168484 | Std - 0.02632716 | Min - 0.6865867 | Max - 0.7532981\n"
     ]
    },
    {
     "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": [
    "t0 = time()\n",
    "gb = GradientBoostingClassifier()\n",
    "gb.fit(x_train, y_train)\n",
    "print(\"Tempo de execução Gradient Boosting:\", round(time()-t0, 3), \"s\")\n",
    "draw_roc_score(gb, x_train, y_train, x_test, y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Modelo Selecionado e Tuning\n",
    "\n",
    "Como pudemos ver na seção anterior, o Gradient Boosting obteve o melhor desempenho utilizando uma instância padrão do classificador, ou seja, sem realizar qualquer ajuste nos seus hiperparâmetros. O proximo passo é identificar o melhor ajuste dos hiperparâmetros para o modelo selecionado, para isso, irei selecionar poucos hiperparâmetros a cada ajuste, afim de reduzir o tempo da busca e para tentar entender como está sendo a importância dada a cada feature. Nos passos a seguir, cada ajuste ideal encontrado será inserido como hiperparâmetro fixo na etapa seguinte.\n",
    "\n",
    "\n",
    "Para iniciar, irei olhar a importância das features com o classificador padrão:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.plotly.v1+json": {
       "data": [
        {
         "marker": {
          "color": [
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           0,
           6.148954295121838e-08,
           0.0005911243930115693,
           0.00061023922746636,
           0.001111993570618346,
           0.001218590863212066,
           0.0012204201416554786,
           0.001253734748009814,
           0.0012778085838116822,
           0.0014696247827685604,
           0.0014771388399899131,
           0.0015257928310974172,
           0.0017585445300000266,
           0.00192969694229352,
           0.001943621198750431,
           0.002015847558841915,
           0.0020741220719496193,
           0.002117014669628676,
           0.0022956979072829074,
           0.00234303551861185,
           0.0023936407776277025,
           0.0025272833554531404,
           0.0026065928282293237,
           0.002767075856036654,
           0.00277824766081995,
           0.0027947322848245353,
           0.0029192058558501688,
           0.00294993568329533,
           0.003034436607213735,
           0.0031584266750483707,
           0.003178762561382551,
           0.003254059921775736,
           0.003368490349036757,
           0.0033822349041226518,
           0.003435085995827429,
           0.0035602415674510428,
           0.003580535634927915,
           0.003617152893064801,
           0.0038654566770736027,
           0.0038941882932635474,
           0.004080626945323943,
           0.004189320008521776,
           0.004193744852504036,
           0.004717534331341,
           0.004964718633039195,
           0.005196598317087741,
           0.005212557123560974,
           0.005386930129493076,
           0.00539839309690986,
           0.005530929621322871,
           0.0057094753917405995,
           0.0059196265671716064,
           0.0060347765952158475,
           0.0060745830593145755,
           0.006106106791825774,
           0.006158423485867479,
           0.006216386877104589,
           0.006324985243685441,
           0.007326991596796609,
           0.008454837780024574,
           0.008653108682554205,
           0.008812322637895532,
           0.010048747407997296,
           0.010646373620178453,
           0.01168086614767548,
           0.01169867787468619,
           0.012040215016277247,
           0.01217194979704381,
           0.012484876016859523,
           0.013032814010348797,
           0.013567861183624248,
           0.0140650852612756,
           0.01697872886694824,
           0.017644748969909318,
           0.01883307297098423,
           0.019899968220826115,
           0.024609676359228585,
           0.02472420068875465,
           0.025514964314461673,
           0.028696299308362806,
           0.02901124081684272,
           0.03240040880740961,
           0.0367845358069597,
           0.03795923378472825,
           0.03831436420162226,
           0.03919805761848015,
           0.13299948467865438,
           0.1330306742306914
          ],
          "colorscale": "Jet",
          "reversescale": true
         },
         "name": "Importância das features",
         "orientation": "h",
         "type": "bar",
         "uid": "301d8e24-edeb-11e8-bfb1-605718f38acb",
         "x": [
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          0,
          6.148954295121838e-08,
          0.0005911243930115693,
          0.00061023922746636,
          0.001111993570618346,
          0.001218590863212066,
          0.0012204201416554786,
          0.001253734748009814,
          0.0012778085838116822,
          0.0014696247827685604,
          0.0014771388399899131,
          0.0015257928310974172,
          0.0017585445300000266,
          0.00192969694229352,
          0.001943621198750431,
          0.002015847558841915,
          0.0020741220719496193,
          0.002117014669628676,
          0.0022956979072829074,
          0.00234303551861185,
          0.0023936407776277025,
          0.0025272833554531404,
          0.0026065928282293237,
          0.002767075856036654,
          0.00277824766081995,
          0.0027947322848245353,
          0.0029192058558501688,
          0.00294993568329533,
          0.003034436607213735,
          0.0031584266750483707,
          0.003178762561382551,
          0.003254059921775736,
          0.003368490349036757,
          0.0033822349041226518,
          0.003435085995827429,
          0.0035602415674510428,
          0.003580535634927915,
          0.003617152893064801,
          0.0038654566770736027,
          0.0038941882932635474,
          0.004080626945323943,
          0.004189320008521776,
          0.004193744852504036,
          0.004717534331341,
          0.004964718633039195,
          0.005196598317087741,
          0.005212557123560974,
          0.005386930129493076,
          0.00539839309690986,
          0.005530929621322871,
          0.0057094753917405995,
          0.0059196265671716064,
          0.0060347765952158475,
          0.0060745830593145755,
          0.006106106791825774,
          0.006158423485867479,
          0.006216386877104589,
          0.006324985243685441,
          0.007326991596796609,
          0.008454837780024574,
          0.008653108682554205,
          0.008812322637895532,
          0.010048747407997296,
          0.010646373620178453,
          0.01168086614767548,
          0.01169867787468619,
          0.012040215016277247,
          0.01217194979704381,
          0.012484876016859523,
          0.013032814010348797,
          0.013567861183624248,
          0.0140650852612756,
          0.01697872886694824,
          0.017644748969909318,
          0.01883307297098423,
          0.019899968220826115,
          0.024609676359228585,
          0.02472420068875465,
          0.025514964314461673,
          0.028696299308362806,
          0.02901124081684272,
          0.03240040880740961,
          0.0367845358069597,
          0.03795923378472825,
          0.03831436420162226,
          0.03919805761848015,
          0.13299948467865438,
          0.1330306742306914
         ],
         "y": [
          "1",
          "AGE",
          "AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY",
          "AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY EXT_SOURCE_2_POLY",
          "AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY EXT_SOURCE_3_POLY",
          "AMT_CREDIT_POLY DAYS_BIRTH_POLY",
          "AMT_CREDIT_POLY DAYS_BIRTH_POLY EXT_SOURCE_2_POLY",
          "AMT_CREDIT_POLY DAYS_BIRTH_POLY EXT_SOURCE_3_POLY",
          "AMT_CREDIT_POLY DAYS_BIRTH_POLY^2",
          "AMT_CREDIT_POLY EXT_SOURCE_2_POLY",
          "AMT_CREDIT_POLY EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY",
          "AMT_CREDIT_POLY EXT_SOURCE_2_POLY^2",
          "AMT_CREDIT_POLY EXT_SOURCE_3_POLY",
          "AMT_CREDIT_POLY EXT_SOURCE_3_POLY^2",
          "AMT_CREDIT_POLY^2 DAYS_BIRTH_POLY",
          "AMT_CREDIT_POLY^2 EXT_SOURCE_2_POLY",
          "AMT_CREDIT_POLY^2 EXT_SOURCE_3_POLY",
          "AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY",
          "AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY EXT_SOURCE_2_POLY",
          "AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY EXT_SOURCE_3_POLY",
          "AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY^2",
          "AMT_INCOME_TOTAL_POLY EXT_SOURCE_2_POLY",
          "AMT_INCOME_TOTAL_POLY EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY",
          "AMT_INCOME_TOTAL_POLY EXT_SOURCE_2_POLY^2",
          "AMT_INCOME_TOTAL_POLY EXT_SOURCE_3_POLY",
          "AMT_INCOME_TOTAL_POLY EXT_SOURCE_3_POLY^2",
          "AMT_INCOME_TOTAL_POLY^2 DAYS_BIRTH_POLY",
          "AMT_INCOME_TOTAL_POLY^2 EXT_SOURCE_2_POLY",
          "AMT_INCOME_TOTAL_POLY^2 EXT_SOURCE_3_POLY",
          "AMT_REQ_CREDIT_BUREAU_DAY",
          "AMT_REQ_CREDIT_BUREAU_HOUR",
          "AMT_REQ_CREDIT_BUREAU_MON",
          "AMT_REQ_CREDIT_BUREAU_QRT",
          "AMT_REQ_CREDIT_BUREAU_WEEK",
          "CNT_CHILDREN",
          "CNT_FAM_MEMBERS",
          "DAYS_BIRTH_POLY",
          "DAYS_BIRTH_POLY EXT_SOURCE_2_POLY",
          "DAYS_BIRTH_POLY EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY",
          "DAYS_BIRTH_POLY EXT_SOURCE_2_POLY^2",
          "DAYS_BIRTH_POLY EXT_SOURCE_3_POLY",
          "DAYS_BIRTH_POLY EXT_SOURCE_3_POLY^2",
          "DAYS_BIRTH_POLY^2",
          "DAYS_BIRTH_POLY^2 EXT_SOURCE_2_POLY",
          "DAYS_BIRTH_POLY^2 EXT_SOURCE_3_POLY",
          "DAYS_BIRTH_POLY^3",
          "DEF_30_CNT_SOCIAL_CIRCLE",
          "DEF_60_CNT_SOCIAL_CIRCLE",
          "EMERGENCYSTATE_MODE_No",
          "EMERGENCYSTATE_MODE_Yes",
          "EMERGENCYSTATE_MODE_nan",
          "EXT_SOURCE_2_POLY",
          "EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY",
          "EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY^2",
          "EXT_SOURCE_2_POLY^2",
          "EXT_SOURCE_2_POLY^2 EXT_SOURCE_3_POLY",
          "EXT_SOURCE_2_POLY^3",
          "EXT_SOURCE_3_POLY",
          "EXT_SOURCE_3_POLY^2",
          "EXT_SOURCE_3_POLY^3",
          "FLAG_CONT_MOBILE",
          "FLAG_DOCUMENT_10",
          "FLAG_DOCUMENT_11",
          "FLAG_DOCUMENT_12",
          "FLAG_DOCUMENT_13",
          "FLAG_DOCUMENT_14",
          "FLAG_DOCUMENT_15",
          "FLAG_DOCUMENT_16",
          "FLAG_DOCUMENT_17",
          "FLAG_DOCUMENT_18",
          "FLAG_DOCUMENT_19",
          "FLAG_DOCUMENT_2",
          "FLAG_DOCUMENT_20",
          "FLAG_DOCUMENT_21",
          "FLAG_DOCUMENT_4",
          "FLAG_DOCUMENT_5",
          "FLAG_DOCUMENT_6",
          "FLAG_DOCUMENT_7",
          "FLAG_DOCUMENT_8",
          "FLAG_DOCUMENT_9",
          "FLAG_EMAIL",
          "FLAG_EMP_PHONE",
          "FLAG_MOBIL",
          "FLAG_OWN_CAR_N",
          "FLAG_OWN_REALTY_Y",
          "FLAG_WORK_PHONE",
          "LIVE_CITY_NOT_WORK_CITY",
          "LIVE_REGION_NOT_WORK_REGION",
          "NAME_EDUCATION_TYPE_Academic degree",
          "NAME_EDUCATION_TYPE_Lower secondary",
          "NAME_FAMILY_STATUS_Separated",
          "NAME_FAMILY_STATUS_Single / not married",
          "NAME_FAMILY_STATUS_Widow",
          "NAME_HOUSING_TYPE_Co-op apartment",
          "NAME_HOUSING_TYPE_House / apartment",
          "NAME_HOUSING_TYPE_Municipal apartment",
          "NAME_INCOME_TYPE_Businessman",
          "NAME_INCOME_TYPE_Commercial associate",
          "NAME_INCOME_TYPE_Pensioner",
          "NAME_INCOME_TYPE_State servant",
          "NAME_INCOME_TYPE_Student",
          "NAME_INCOME_TYPE_Unemployed",
          "NAME_TYPE_SUITE_Other_A",
          "NAME_TYPE_SUITE_Other_B",
          "NAME_TYPE_SUITE_Unaccompanied",
          "OCCUPATION_TYPE_Accountants",
          "OCCUPATION_TYPE_Cleaning staff",
          "OCCUPATION_TYPE_High skill tech staff",
          "OCCUPATION_TYPE_IT staff",
          "OCCUPATION_TYPE_Managers",
          "OCCUPATION_TYPE_Medicine staff",
          "OCCUPATION_TYPE_Private service staff",
          "OCCUPATION_TYPE_Realty agents",
          "OCCUPATION_TYPE_Secretaries",
          "ORGANIZATION_TYPE_Advertising",
          "ORGANIZATION_TYPE_Agriculture",
          "ORGANIZATION_TYPE_Business Entity Type 1",
          "ORGANIZATION_TYPE_Culture",
          "ORGANIZATION_TYPE_Emergency",
          "ORGANIZATION_TYPE_Government",
          "ORGANIZATION_TYPE_Hotel",
          "ORGANIZATION_TYPE_Industry: type 1",
          "ORGANIZATION_TYPE_Industry: type 10",
          "ORGANIZATION_TYPE_Industry: type 11",
          "ORGANIZATION_TYPE_Industry: type 12",
          "ORGANIZATION_TYPE_Industry: type 3",
          "ORGANIZATION_TYPE_Industry: type 4",
          "ORGANIZATION_TYPE_Industry: type 5",
          "ORGANIZATION_TYPE_Industry: type 6",
          "ORGANIZATION_TYPE_Industry: type 7",
          "ORGANIZATION_TYPE_Industry: type 8",
          "ORGANIZATION_TYPE_Industry: type 9",
          "ORGANIZATION_TYPE_Insurance",
          "ORGANIZATION_TYPE_Medicine",
          "ORGANIZATION_TYPE_Military",
          "ORGANIZATION_TYPE_Mobile",
          "ORGANIZATION_TYPE_Other",
          "ORGANIZATION_TYPE_Police",
          "ORGANIZATION_TYPE_Postal",
          "ORGANIZATION_TYPE_Realtor",
          "ORGANIZATION_TYPE_Religion",
          "ORGANIZATION_TYPE_School",
          "ORGANIZATION_TYPE_Security",
          "ORGANIZATION_TYPE_Security Ministries",
          "ORGANIZATION_TYPE_Services",
          "ORGANIZATION_TYPE_Telecom",
          "ORGANIZATION_TYPE_Trade: type 1",
          "ORGANIZATION_TYPE_Trade: type 2",
          "ORGANIZATION_TYPE_Trade: type 3",
          "ORGANIZATION_TYPE_Trade: type 4",
          "ORGANIZATION_TYPE_Trade: type 5",
          "ORGANIZATION_TYPE_Trade: type 6",
          "ORGANIZATION_TYPE_Trade: type 7",
          "ORGANIZATION_TYPE_Transport: type 1",
          "ORGANIZATION_TYPE_Transport: type 2",
          "ORGANIZATION_TYPE_Transport: type 4",
          "ORGANIZATION_TYPE_University",
          "ORGANIZATION_TYPE_XNA",
          "REGION_POPULATION_RELATIVE",
          "REGION_RATING_CLIENT",
          "REGION_RATING_CLIENT_W_CITY",
          "REG_CITY_NOT_LIVE_CITY",
          "REG_CITY_NOT_WORK_CITY",
          "REG_REGION_NOT_LIVE_REGION",
          "REG_REGION_NOT_WORK_REGION",
          "WEEKDAY_APPR_PROCESS_START_MONDAY",
          "WEEKDAY_APPR_PROCESS_START_SATURDAY",
          "WEEKDAY_APPR_PROCESS_START_SUNDAY",
          "WEEKDAY_APPR_PROCESS_START_THURSDAY",
          "YEAR_EMPLOYED",
          "NAME_CONTRACT_TYPE_Cash loans",
          "ORGANIZATION_TYPE_Bank",
          "ORGANIZATION_TYPE_Business Entity Type 3",
          "ORGANIZATION_TYPE_Restaurant",
          "AMT_CREDIT_LOG",
          "ORGANIZATION_TYPE_Housing",
          "ORGANIZATION_TYPE_Self-employed",
          "ORGANIZATION_TYPE_Construction",
          "OCCUPATION_TYPE_HR staff",
          "WEEKDAY_APPR_PROCESS_START_FRIDAY",
          "ORGANIZATION_TYPE_Business Entity Type 2",
          "AMT_INCOME_TOTAL_PER_MEMBER",
          "AMT_INCOME_TOTAL_POLY^3",
          "NAME_FAMILY_STATUS_Civil marriage",
          "WEEKDAY_APPR_PROCESS_START_WEDNESDAY",
          "NAME_TYPE_SUITE_Spouse, partner",
          "OCCUPATION_TYPE_Core staff",
          "NAME_EDUCATION_TYPE_Incomplete higher",
          "NAME_HOUSING_TYPE_Office apartment",
          "ORGANIZATION_TYPE_Electricity",
          "AMT_INCOME_TOTAL_POLY",
          "ORGANIZATION_TYPE_Industry: type 2",
          "OCCUPATION_TYPE_Security staff",
          "YEAR_REGISTRATION",
          "NAME_TYPE_SUITE_nan",
          "OCCUPATION_TYPE_Waiters/barmen staff",
          "WEEKDAY_APPR_PROCESS_START_TUESDAY",
          "NAME_CONTRACT_TYPE_Revolving loans",
          "OCCUPATION_TYPE_Sales staff",
          "OCCUPATION_TYPE_nan",
          "AMT_INCOME_TOTAL_LOG",
          "NAME_TYPE_SUITE_Group of people",
          "ORGANIZATION_TYPE_Cleaning",
          "OCCUPATION_TYPE_Drivers",
          "NAME_TYPE_SUITE_Children",
          "ORGANIZATION_TYPE_Kindergarten",
          "AMT_INCOME_TOTAL_POLY^2",
          "ORGANIZATION_TYPE_Legal Services",
          "AMT_ANNUITY_LOG",
          "NAME_INCOME_TYPE_Working",
          "ORGANIZATION_TYPE_Transport: type 3",
          "NAME_HOUSING_TYPE_Rented apartment",
          "FLAG_DOCUMENT_3",
          "OBS_30_CNT_SOCIAL_CIRCLE",
          "FLAG_OWN_REALTY_N",
          "YEARS_BEGINEXPLUATATION_AVG",
          "FLAG_PHONE",
          "NAME_EDUCATION_TYPE_Higher education",
          "OCCUPATION_TYPE_Cooking staff",
          "OCCUPATION_TYPE_Laborers",
          "OBS_60_CNT_SOCIAL_CIRCLE",
          "NAME_TYPE_SUITE_Family",
          "AMT_CREDIT",
          "FLAG_OWN_CAR_Y",
          "NAME_EDUCATION_TYPE_Secondary / secondary special",
          "NAME_HOUSING_TYPE_With parents",
          "NAME_FAMILY_STATUS_Married",
          "AMT_REQ_CREDIT_BUREAU_YEAR",
          "ORGANIZATION_TYPE_Industry: type 13",
          "CODE_GENDER_F",
          "AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY",
          "YEARS_BEGINEXPLUATATION_MODE",
          "AMT_INCOME_TOTAL",
          "AMT_CREDIT_POLY",
          "AMT_GOODS_PRICE",
          "DAYS_BIRTH",
          "AMT_ANNUITY",
          "HOUR_APPR_PROCESS_START",
          "CODE_GENDER_M",
          "OCCUPATION_TYPE_Low-skill Laborers",
          "AMT_CREDIT_POLY^3",
          "DAYS_LAST_PHONE_CHANGE",
          "AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY^2",
          "CREDIT_PER_MEMBER",
          "AMT_CREDIT_PER_AGE",
          "AMT_CREDIT_POLY^2",
          "AMT_INCOME_TOTAL_PER_AGE",
          "DAYS_REGISTRATION",
          "CREDIT_PER_INCOME",
          "YEARS_BEGINEXPLUATATION_MEDI",
          "AMT_CREDIT_PER_ANNUITY",
          "AMT_CREDIT_POLY^2 AMT_INCOME_TOTAL_POLY",
          "DAYS_EMPLOYED",
          "ANNUITY_PER_INCOME",
          "DAYS_ID_PUBLISH",
          "EXT_SOURCE_2",
          "EXT_SOURCE_3"
         ]
        }
       ],
       "layout": {
        "height": 4000,
        "margin": {
         "l": 300
        },
        "title": "Importância das features",
        "width": 900,
        "yaxis": {
         "showgrid": true
        }
       }
      },
      "text/html": [
       "<div id=\"8036825f-95c5-43d1-af2b-ddc520e77848\" style=\"height: 4000px; width: 900px;\" class=\"plotly-graph-div\"></div><script type=\"text/javascript\">require([\"plotly\"], function(Plotly) { window.PLOTLYENV=window.PLOTLYENV || {};window.PLOTLYENV.BASE_URL=\"https://plot.ly\";Plotly.newPlot(\"8036825f-95c5-43d1-af2b-ddc520e77848\", [{\"marker\": {\"color\": [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 6.148954295121838e-08, 0.0005911243930115693, 0.00061023922746636, 0.001111993570618346, 0.001218590863212066, 0.0012204201416554786, 0.001253734748009814, 0.0012778085838116822, 0.0014696247827685604, 0.0014771388399899131, 0.0015257928310974172, 0.0017585445300000266, 0.00192969694229352, 0.001943621198750431, 0.002015847558841915, 0.0020741220719496193, 0.002117014669628676, 0.0022956979072829074, 0.00234303551861185, 0.0023936407776277025, 0.0025272833554531404, 0.0026065928282293237, 0.002767075856036654, 0.00277824766081995, 0.0027947322848245353, 0.0029192058558501688, 0.00294993568329533, 0.003034436607213735, 0.0031584266750483707, 0.003178762561382551, 0.003254059921775736, 0.003368490349036757, 0.0033822349041226518, 0.003435085995827429, 0.0035602415674510428, 0.003580535634927915, 0.003617152893064801, 0.0038654566770736027, 0.0038941882932635474, 0.004080626945323943, 0.004189320008521776, 0.004193744852504036, 0.004717534331341, 0.004964718633039195, 0.005196598317087741, 0.005212557123560974, 0.005386930129493076, 0.00539839309690986, 0.005530929621322871, 0.0057094753917405995, 0.0059196265671716064, 0.0060347765952158475, 0.0060745830593145755, 0.006106106791825774, 0.006158423485867479, 0.006216386877104589, 0.006324985243685441, 0.007326991596796609, 0.008454837780024574, 0.008653108682554205, 0.008812322637895532, 0.010048747407997296, 0.010646373620178453, 0.01168086614767548, 0.01169867787468619, 0.012040215016277247, 0.01217194979704381, 0.012484876016859523, 0.013032814010348797, 0.013567861183624248, 0.0140650852612756, 0.01697872886694824, 0.017644748969909318, 0.01883307297098423, 0.019899968220826115, 0.024609676359228585, 0.02472420068875465, 0.025514964314461673, 0.028696299308362806, 0.02901124081684272, 0.03240040880740961, 0.0367845358069597, 0.03795923378472825, 0.03831436420162226, 0.03919805761848015, 0.13299948467865438, 0.1330306742306914], \"colorscale\": \"Jet\", \"reversescale\": true}, \"name\": \"Import\\u00e2ncia das features\", \"orientation\": \"h\", \"x\": [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 6.148954295121838e-08, 0.0005911243930115693, 0.00061023922746636, 0.001111993570618346, 0.001218590863212066, 0.0012204201416554786, 0.001253734748009814, 0.0012778085838116822, 0.0014696247827685604, 0.0014771388399899131, 0.0015257928310974172, 0.0017585445300000266, 0.00192969694229352, 0.001943621198750431, 0.002015847558841915, 0.0020741220719496193, 0.002117014669628676, 0.0022956979072829074, 0.00234303551861185, 0.0023936407776277025, 0.0025272833554531404, 0.0026065928282293237, 0.002767075856036654, 0.00277824766081995, 0.0027947322848245353, 0.0029192058558501688, 0.00294993568329533, 0.003034436607213735, 0.0031584266750483707, 0.003178762561382551, 0.003254059921775736, 0.003368490349036757, 0.0033822349041226518, 0.003435085995827429, 0.0035602415674510428, 0.003580535634927915, 0.003617152893064801, 0.0038654566770736027, 0.0038941882932635474, 0.004080626945323943, 0.004189320008521776, 0.004193744852504036, 0.004717534331341, 0.004964718633039195, 0.005196598317087741, 0.005212557123560974, 0.005386930129493076, 0.00539839309690986, 0.005530929621322871, 0.0057094753917405995, 0.0059196265671716064, 0.0060347765952158475, 0.0060745830593145755, 0.006106106791825774, 0.006158423485867479, 0.006216386877104589, 0.006324985243685441, 0.007326991596796609, 0.008454837780024574, 0.008653108682554205, 0.008812322637895532, 0.010048747407997296, 0.010646373620178453, 0.01168086614767548, 0.01169867787468619, 0.012040215016277247, 0.01217194979704381, 0.012484876016859523, 0.013032814010348797, 0.013567861183624248, 0.0140650852612756, 0.01697872886694824, 0.017644748969909318, 0.01883307297098423, 0.019899968220826115, 0.024609676359228585, 0.02472420068875465, 0.025514964314461673, 0.028696299308362806, 0.02901124081684272, 0.03240040880740961, 0.0367845358069597, 0.03795923378472825, 0.03831436420162226, 0.03919805761848015, 0.13299948467865438, 0.1330306742306914], \"y\": [\"1\", \"AGE\", \"AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY\", \"AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY EXT_SOURCE_2_POLY\", \"AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY EXT_SOURCE_3_POLY\", \"AMT_CREDIT_POLY DAYS_BIRTH_POLY\", \"AMT_CREDIT_POLY DAYS_BIRTH_POLY EXT_SOURCE_2_POLY\", \"AMT_CREDIT_POLY DAYS_BIRTH_POLY EXT_SOURCE_3_POLY\", \"AMT_CREDIT_POLY DAYS_BIRTH_POLY^2\", \"AMT_CREDIT_POLY EXT_SOURCE_2_POLY\", \"AMT_CREDIT_POLY EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY\", \"AMT_CREDIT_POLY EXT_SOURCE_2_POLY^2\", \"AMT_CREDIT_POLY EXT_SOURCE_3_POLY\", \"AMT_CREDIT_POLY EXT_SOURCE_3_POLY^2\", \"AMT_CREDIT_POLY^2 DAYS_BIRTH_POLY\", \"AMT_CREDIT_POLY^2 EXT_SOURCE_2_POLY\", \"AMT_CREDIT_POLY^2 EXT_SOURCE_3_POLY\", \"AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY\", \"AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY EXT_SOURCE_2_POLY\", \"AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY EXT_SOURCE_3_POLY\", \"AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY^2\", \"AMT_INCOME_TOTAL_POLY EXT_SOURCE_2_POLY\", \"AMT_INCOME_TOTAL_POLY EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY\", \"AMT_INCOME_TOTAL_POLY EXT_SOURCE_2_POLY^2\", \"AMT_INCOME_TOTAL_POLY EXT_SOURCE_3_POLY\", \"AMT_INCOME_TOTAL_POLY EXT_SOURCE_3_POLY^2\", \"AMT_INCOME_TOTAL_POLY^2 DAYS_BIRTH_POLY\", \"AMT_INCOME_TOTAL_POLY^2 EXT_SOURCE_2_POLY\", \"AMT_INCOME_TOTAL_POLY^2 EXT_SOURCE_3_POLY\", \"AMT_REQ_CREDIT_BUREAU_DAY\", \"AMT_REQ_CREDIT_BUREAU_HOUR\", \"AMT_REQ_CREDIT_BUREAU_MON\", \"AMT_REQ_CREDIT_BUREAU_QRT\", \"AMT_REQ_CREDIT_BUREAU_WEEK\", \"CNT_CHILDREN\", \"CNT_FAM_MEMBERS\", \"DAYS_BIRTH_POLY\", \"DAYS_BIRTH_POLY EXT_SOURCE_2_POLY\", \"DAYS_BIRTH_POLY EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY\", \"DAYS_BIRTH_POLY EXT_SOURCE_2_POLY^2\", \"DAYS_BIRTH_POLY EXT_SOURCE_3_POLY\", \"DAYS_BIRTH_POLY EXT_SOURCE_3_POLY^2\", \"DAYS_BIRTH_POLY^2\", \"DAYS_BIRTH_POLY^2 EXT_SOURCE_2_POLY\", \"DAYS_BIRTH_POLY^2 EXT_SOURCE_3_POLY\", \"DAYS_BIRTH_POLY^3\", \"DEF_30_CNT_SOCIAL_CIRCLE\", \"DEF_60_CNT_SOCIAL_CIRCLE\", \"EMERGENCYSTATE_MODE_No\", \"EMERGENCYSTATE_MODE_Yes\", \"EMERGENCYSTATE_MODE_nan\", \"EXT_SOURCE_2_POLY\", \"EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY\", \"EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY^2\", \"EXT_SOURCE_2_POLY^2\", \"EXT_SOURCE_2_POLY^2 EXT_SOURCE_3_POLY\", \"EXT_SOURCE_2_POLY^3\", \"EXT_SOURCE_3_POLY\", \"EXT_SOURCE_3_POLY^2\", \"EXT_SOURCE_3_POLY^3\", \"FLAG_CONT_MOBILE\", \"FLAG_DOCUMENT_10\", \"FLAG_DOCUMENT_11\", \"FLAG_DOCUMENT_12\", \"FLAG_DOCUMENT_13\", \"FLAG_DOCUMENT_14\", \"FLAG_DOCUMENT_15\", \"FLAG_DOCUMENT_16\", \"FLAG_DOCUMENT_17\", \"FLAG_DOCUMENT_18\", \"FLAG_DOCUMENT_19\", \"FLAG_DOCUMENT_2\", \"FLAG_DOCUMENT_20\", \"FLAG_DOCUMENT_21\", \"FLAG_DOCUMENT_4\", \"FLAG_DOCUMENT_5\", \"FLAG_DOCUMENT_6\", \"FLAG_DOCUMENT_7\", \"FLAG_DOCUMENT_8\", \"FLAG_DOCUMENT_9\", \"FLAG_EMAIL\", \"FLAG_EMP_PHONE\", \"FLAG_MOBIL\", \"FLAG_OWN_CAR_N\", \"FLAG_OWN_REALTY_Y\", \"FLAG_WORK_PHONE\", \"LIVE_CITY_NOT_WORK_CITY\", \"LIVE_REGION_NOT_WORK_REGION\", \"NAME_EDUCATION_TYPE_Academic degree\", \"NAME_EDUCATION_TYPE_Lower secondary\", \"NAME_FAMILY_STATUS_Separated\", \"NAME_FAMILY_STATUS_Single / not married\", \"NAME_FAMILY_STATUS_Widow\", \"NAME_HOUSING_TYPE_Co-op apartment\", \"NAME_HOUSING_TYPE_House / apartment\", \"NAME_HOUSING_TYPE_Municipal apartment\", \"NAME_INCOME_TYPE_Businessman\", \"NAME_INCOME_TYPE_Commercial associate\", \"NAME_INCOME_TYPE_Pensioner\", \"NAME_INCOME_TYPE_State servant\", \"NAME_INCOME_TYPE_Student\", \"NAME_INCOME_TYPE_Unemployed\", \"NAME_TYPE_SUITE_Other_A\", \"NAME_TYPE_SUITE_Other_B\", \"NAME_TYPE_SUITE_Unaccompanied\", \"OCCUPATION_TYPE_Accountants\", \"OCCUPATION_TYPE_Cleaning staff\", \"OCCUPATION_TYPE_High skill tech staff\", \"OCCUPATION_TYPE_IT staff\", \"OCCUPATION_TYPE_Managers\", \"OCCUPATION_TYPE_Medicine staff\", \"OCCUPATION_TYPE_Private service staff\", \"OCCUPATION_TYPE_Realty agents\", \"OCCUPATION_TYPE_Secretaries\", \"ORGANIZATION_TYPE_Advertising\", \"ORGANIZATION_TYPE_Agriculture\", \"ORGANIZATION_TYPE_Business Entity Type 1\", \"ORGANIZATION_TYPE_Culture\", \"ORGANIZATION_TYPE_Emergency\", \"ORGANIZATION_TYPE_Government\", \"ORGANIZATION_TYPE_Hotel\", \"ORGANIZATION_TYPE_Industry: type 1\", \"ORGANIZATION_TYPE_Industry: type 10\", \"ORGANIZATION_TYPE_Industry: type 11\", \"ORGANIZATION_TYPE_Industry: type 12\", \"ORGANIZATION_TYPE_Industry: type 3\", \"ORGANIZATION_TYPE_Industry: type 4\", \"ORGANIZATION_TYPE_Industry: type 5\", \"ORGANIZATION_TYPE_Industry: type 6\", \"ORGANIZATION_TYPE_Industry: type 7\", \"ORGANIZATION_TYPE_Industry: type 8\", \"ORGANIZATION_TYPE_Industry: type 9\", \"ORGANIZATION_TYPE_Insurance\", \"ORGANIZATION_TYPE_Medicine\", \"ORGANIZATION_TYPE_Military\", \"ORGANIZATION_TYPE_Mobile\", \"ORGANIZATION_TYPE_Other\", \"ORGANIZATION_TYPE_Police\", \"ORGANIZATION_TYPE_Postal\", \"ORGANIZATION_TYPE_Realtor\", \"ORGANIZATION_TYPE_Religion\", \"ORGANIZATION_TYPE_School\", \"ORGANIZATION_TYPE_Security\", \"ORGANIZATION_TYPE_Security Ministries\", \"ORGANIZATION_TYPE_Services\", \"ORGANIZATION_TYPE_Telecom\", \"ORGANIZATION_TYPE_Trade: type 1\", \"ORGANIZATION_TYPE_Trade: type 2\", \"ORGANIZATION_TYPE_Trade: type 3\", \"ORGANIZATION_TYPE_Trade: type 4\", \"ORGANIZATION_TYPE_Trade: type 5\", \"ORGANIZATION_TYPE_Trade: type 6\", \"ORGANIZATION_TYPE_Trade: type 7\", \"ORGANIZATION_TYPE_Transport: type 1\", \"ORGANIZATION_TYPE_Transport: type 2\", \"ORGANIZATION_TYPE_Transport: type 4\", \"ORGANIZATION_TYPE_University\", \"ORGANIZATION_TYPE_XNA\", \"REGION_POPULATION_RELATIVE\", \"REGION_RATING_CLIENT\", \"REGION_RATING_CLIENT_W_CITY\", \"REG_CITY_NOT_LIVE_CITY\", \"REG_CITY_NOT_WORK_CITY\", \"REG_REGION_NOT_LIVE_REGION\", \"REG_REGION_NOT_WORK_REGION\", \"WEEKDAY_APPR_PROCESS_START_MONDAY\", \"WEEKDAY_APPR_PROCESS_START_SATURDAY\", \"WEEKDAY_APPR_PROCESS_START_SUNDAY\", \"WEEKDAY_APPR_PROCESS_START_THURSDAY\", \"YEAR_EMPLOYED\", \"NAME_CONTRACT_TYPE_Cash loans\", \"ORGANIZATION_TYPE_Bank\", \"ORGANIZATION_TYPE_Business Entity Type 3\", \"ORGANIZATION_TYPE_Restaurant\", \"AMT_CREDIT_LOG\", \"ORGANIZATION_TYPE_Housing\", \"ORGANIZATION_TYPE_Self-employed\", \"ORGANIZATION_TYPE_Construction\", \"OCCUPATION_TYPE_HR staff\", \"WEEKDAY_APPR_PROCESS_START_FRIDAY\", \"ORGANIZATION_TYPE_Business Entity Type 2\", \"AMT_INCOME_TOTAL_PER_MEMBER\", \"AMT_INCOME_TOTAL_POLY^3\", \"NAME_FAMILY_STATUS_Civil marriage\", \"WEEKDAY_APPR_PROCESS_START_WEDNESDAY\", \"NAME_TYPE_SUITE_Spouse, partner\", \"OCCUPATION_TYPE_Core staff\", \"NAME_EDUCATION_TYPE_Incomplete higher\", \"NAME_HOUSING_TYPE_Office apartment\", \"ORGANIZATION_TYPE_Electricity\", \"AMT_INCOME_TOTAL_POLY\", \"ORGANIZATION_TYPE_Industry: type 2\", \"OCCUPATION_TYPE_Security staff\", \"YEAR_REGISTRATION\", \"NAME_TYPE_SUITE_nan\", \"OCCUPATION_TYPE_Waiters/barmen staff\", \"WEEKDAY_APPR_PROCESS_START_TUESDAY\", \"NAME_CONTRACT_TYPE_Revolving loans\", \"OCCUPATION_TYPE_Sales staff\", \"OCCUPATION_TYPE_nan\", \"AMT_INCOME_TOTAL_LOG\", \"NAME_TYPE_SUITE_Group of people\", \"ORGANIZATION_TYPE_Cleaning\", \"OCCUPATION_TYPE_Drivers\", \"NAME_TYPE_SUITE_Children\", \"ORGANIZATION_TYPE_Kindergarten\", \"AMT_INCOME_TOTAL_POLY^2\", \"ORGANIZATION_TYPE_Legal Services\", \"AMT_ANNUITY_LOG\", \"NAME_INCOME_TYPE_Working\", \"ORGANIZATION_TYPE_Transport: type 3\", \"NAME_HOUSING_TYPE_Rented apartment\", \"FLAG_DOCUMENT_3\", \"OBS_30_CNT_SOCIAL_CIRCLE\", \"FLAG_OWN_REALTY_N\", \"YEARS_BEGINEXPLUATATION_AVG\", \"FLAG_PHONE\", \"NAME_EDUCATION_TYPE_Higher education\", \"OCCUPATION_TYPE_Cooking staff\", \"OCCUPATION_TYPE_Laborers\", \"OBS_60_CNT_SOCIAL_CIRCLE\", \"NAME_TYPE_SUITE_Family\", \"AMT_CREDIT\", \"FLAG_OWN_CAR_Y\", \"NAME_EDUCATION_TYPE_Secondary / secondary special\", \"NAME_HOUSING_TYPE_With parents\", \"NAME_FAMILY_STATUS_Married\", \"AMT_REQ_CREDIT_BUREAU_YEAR\", \"ORGANIZATION_TYPE_Industry: type 13\", \"CODE_GENDER_F\", \"AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY\", \"YEARS_BEGINEXPLUATATION_MODE\", \"AMT_INCOME_TOTAL\", \"AMT_CREDIT_POLY\", \"AMT_GOODS_PRICE\", \"DAYS_BIRTH\", \"AMT_ANNUITY\", \"HOUR_APPR_PROCESS_START\", \"CODE_GENDER_M\", \"OCCUPATION_TYPE_Low-skill Laborers\", \"AMT_CREDIT_POLY^3\", \"DAYS_LAST_PHONE_CHANGE\", \"AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY^2\", \"CREDIT_PER_MEMBER\", \"AMT_CREDIT_PER_AGE\", \"AMT_CREDIT_POLY^2\", \"AMT_INCOME_TOTAL_PER_AGE\", \"DAYS_REGISTRATION\", \"CREDIT_PER_INCOME\", \"YEARS_BEGINEXPLUATATION_MEDI\", \"AMT_CREDIT_PER_ANNUITY\", \"AMT_CREDIT_POLY^2 AMT_INCOME_TOTAL_POLY\", \"DAYS_EMPLOYED\", \"ANNUITY_PER_INCOME\", \"DAYS_ID_PUBLISH\", \"EXT_SOURCE_2\", \"EXT_SOURCE_3\"], \"type\": \"bar\", \"uid\": \"301d8e24-edeb-11e8-bfb1-605718f38acb\"}], {\"height\": 4000, \"margin\": {\"l\": 300}, \"title\": \"Import\\u00e2ncia das features\", \"width\": 900, \"yaxis\": {\"showgrid\": true}}, {\"showLink\": true, \"linkText\": \"Export to plot.ly\"})});</script>"
      ],
      "text/vnd.plotly.v1+html": [
       "<div id=\"8036825f-95c5-43d1-af2b-ddc520e77848\" style=\"height: 4000px; width: 900px;\" class=\"plotly-graph-div\"></div><script type=\"text/javascript\">require([\"plotly\"], function(Plotly) { window.PLOTLYENV=window.PLOTLYENV || {};window.PLOTLYENV.BASE_URL=\"https://plot.ly\";Plotly.newPlot(\"8036825f-95c5-43d1-af2b-ddc520e77848\", [{\"marker\": {\"color\": [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 6.148954295121838e-08, 0.0005911243930115693, 0.00061023922746636, 0.001111993570618346, 0.001218590863212066, 0.0012204201416554786, 0.001253734748009814, 0.0012778085838116822, 0.0014696247827685604, 0.0014771388399899131, 0.0015257928310974172, 0.0017585445300000266, 0.00192969694229352, 0.001943621198750431, 0.002015847558841915, 0.0020741220719496193, 0.002117014669628676, 0.0022956979072829074, 0.00234303551861185, 0.0023936407776277025, 0.0025272833554531404, 0.0026065928282293237, 0.002767075856036654, 0.00277824766081995, 0.0027947322848245353, 0.0029192058558501688, 0.00294993568329533, 0.003034436607213735, 0.0031584266750483707, 0.003178762561382551, 0.003254059921775736, 0.003368490349036757, 0.0033822349041226518, 0.003435085995827429, 0.0035602415674510428, 0.003580535634927915, 0.003617152893064801, 0.0038654566770736027, 0.0038941882932635474, 0.004080626945323943, 0.004189320008521776, 0.004193744852504036, 0.004717534331341, 0.004964718633039195, 0.005196598317087741, 0.005212557123560974, 0.005386930129493076, 0.00539839309690986, 0.005530929621322871, 0.0057094753917405995, 0.0059196265671716064, 0.0060347765952158475, 0.0060745830593145755, 0.006106106791825774, 0.006158423485867479, 0.006216386877104589, 0.006324985243685441, 0.007326991596796609, 0.008454837780024574, 0.008653108682554205, 0.008812322637895532, 0.010048747407997296, 0.010646373620178453, 0.01168086614767548, 0.01169867787468619, 0.012040215016277247, 0.01217194979704381, 0.012484876016859523, 0.013032814010348797, 0.013567861183624248, 0.0140650852612756, 0.01697872886694824, 0.017644748969909318, 0.01883307297098423, 0.019899968220826115, 0.024609676359228585, 0.02472420068875465, 0.025514964314461673, 0.028696299308362806, 0.02901124081684272, 0.03240040880740961, 0.0367845358069597, 0.03795923378472825, 0.03831436420162226, 0.03919805761848015, 0.13299948467865438, 0.1330306742306914], \"colorscale\": \"Jet\", \"reversescale\": true}, \"name\": \"Import\\u00e2ncia das features\", \"orientation\": \"h\", \"x\": [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 6.148954295121838e-08, 0.0005911243930115693, 0.00061023922746636, 0.001111993570618346, 0.001218590863212066, 0.0012204201416554786, 0.001253734748009814, 0.0012778085838116822, 0.0014696247827685604, 0.0014771388399899131, 0.0015257928310974172, 0.0017585445300000266, 0.00192969694229352, 0.001943621198750431, 0.002015847558841915, 0.0020741220719496193, 0.002117014669628676, 0.0022956979072829074, 0.00234303551861185, 0.0023936407776277025, 0.0025272833554531404, 0.0026065928282293237, 0.002767075856036654, 0.00277824766081995, 0.0027947322848245353, 0.0029192058558501688, 0.00294993568329533, 0.003034436607213735, 0.0031584266750483707, 0.003178762561382551, 0.003254059921775736, 0.003368490349036757, 0.0033822349041226518, 0.003435085995827429, 0.0035602415674510428, 0.003580535634927915, 0.003617152893064801, 0.0038654566770736027, 0.0038941882932635474, 0.004080626945323943, 0.004189320008521776, 0.004193744852504036, 0.004717534331341, 0.004964718633039195, 0.005196598317087741, 0.005212557123560974, 0.005386930129493076, 0.00539839309690986, 0.005530929621322871, 0.0057094753917405995, 0.0059196265671716064, 0.0060347765952158475, 0.0060745830593145755, 0.006106106791825774, 0.006158423485867479, 0.006216386877104589, 0.006324985243685441, 0.007326991596796609, 0.008454837780024574, 0.008653108682554205, 0.008812322637895532, 0.010048747407997296, 0.010646373620178453, 0.01168086614767548, 0.01169867787468619, 0.012040215016277247, 0.01217194979704381, 0.012484876016859523, 0.013032814010348797, 0.013567861183624248, 0.0140650852612756, 0.01697872886694824, 0.017644748969909318, 0.01883307297098423, 0.019899968220826115, 0.024609676359228585, 0.02472420068875465, 0.025514964314461673, 0.028696299308362806, 0.02901124081684272, 0.03240040880740961, 0.0367845358069597, 0.03795923378472825, 0.03831436420162226, 0.03919805761848015, 0.13299948467865438, 0.1330306742306914], \"y\": [\"1\", \"AGE\", \"AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY\", \"AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY EXT_SOURCE_2_POLY\", \"AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY EXT_SOURCE_3_POLY\", \"AMT_CREDIT_POLY DAYS_BIRTH_POLY\", \"AMT_CREDIT_POLY DAYS_BIRTH_POLY EXT_SOURCE_2_POLY\", \"AMT_CREDIT_POLY DAYS_BIRTH_POLY EXT_SOURCE_3_POLY\", \"AMT_CREDIT_POLY DAYS_BIRTH_POLY^2\", \"AMT_CREDIT_POLY EXT_SOURCE_2_POLY\", \"AMT_CREDIT_POLY EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY\", \"AMT_CREDIT_POLY EXT_SOURCE_2_POLY^2\", \"AMT_CREDIT_POLY EXT_SOURCE_3_POLY\", \"AMT_CREDIT_POLY EXT_SOURCE_3_POLY^2\", \"AMT_CREDIT_POLY^2 DAYS_BIRTH_POLY\", \"AMT_CREDIT_POLY^2 EXT_SOURCE_2_POLY\", \"AMT_CREDIT_POLY^2 EXT_SOURCE_3_POLY\", \"AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY\", \"AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY EXT_SOURCE_2_POLY\", \"AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY EXT_SOURCE_3_POLY\", \"AMT_INCOME_TOTAL_POLY DAYS_BIRTH_POLY^2\", \"AMT_INCOME_TOTAL_POLY EXT_SOURCE_2_POLY\", \"AMT_INCOME_TOTAL_POLY EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY\", \"AMT_INCOME_TOTAL_POLY EXT_SOURCE_2_POLY^2\", \"AMT_INCOME_TOTAL_POLY EXT_SOURCE_3_POLY\", \"AMT_INCOME_TOTAL_POLY EXT_SOURCE_3_POLY^2\", \"AMT_INCOME_TOTAL_POLY^2 DAYS_BIRTH_POLY\", \"AMT_INCOME_TOTAL_POLY^2 EXT_SOURCE_2_POLY\", \"AMT_INCOME_TOTAL_POLY^2 EXT_SOURCE_3_POLY\", \"AMT_REQ_CREDIT_BUREAU_DAY\", \"AMT_REQ_CREDIT_BUREAU_HOUR\", \"AMT_REQ_CREDIT_BUREAU_MON\", \"AMT_REQ_CREDIT_BUREAU_QRT\", \"AMT_REQ_CREDIT_BUREAU_WEEK\", \"CNT_CHILDREN\", \"CNT_FAM_MEMBERS\", \"DAYS_BIRTH_POLY\", \"DAYS_BIRTH_POLY EXT_SOURCE_2_POLY\", \"DAYS_BIRTH_POLY EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY\", \"DAYS_BIRTH_POLY EXT_SOURCE_2_POLY^2\", \"DAYS_BIRTH_POLY EXT_SOURCE_3_POLY\", \"DAYS_BIRTH_POLY EXT_SOURCE_3_POLY^2\", \"DAYS_BIRTH_POLY^2\", \"DAYS_BIRTH_POLY^2 EXT_SOURCE_2_POLY\", \"DAYS_BIRTH_POLY^2 EXT_SOURCE_3_POLY\", \"DAYS_BIRTH_POLY^3\", \"DEF_30_CNT_SOCIAL_CIRCLE\", \"DEF_60_CNT_SOCIAL_CIRCLE\", \"EMERGENCYSTATE_MODE_No\", \"EMERGENCYSTATE_MODE_Yes\", \"EMERGENCYSTATE_MODE_nan\", \"EXT_SOURCE_2_POLY\", \"EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY\", \"EXT_SOURCE_2_POLY EXT_SOURCE_3_POLY^2\", \"EXT_SOURCE_2_POLY^2\", \"EXT_SOURCE_2_POLY^2 EXT_SOURCE_3_POLY\", \"EXT_SOURCE_2_POLY^3\", \"EXT_SOURCE_3_POLY\", \"EXT_SOURCE_3_POLY^2\", \"EXT_SOURCE_3_POLY^3\", \"FLAG_CONT_MOBILE\", \"FLAG_DOCUMENT_10\", \"FLAG_DOCUMENT_11\", \"FLAG_DOCUMENT_12\", \"FLAG_DOCUMENT_13\", \"FLAG_DOCUMENT_14\", \"FLAG_DOCUMENT_15\", \"FLAG_DOCUMENT_16\", \"FLAG_DOCUMENT_17\", \"FLAG_DOCUMENT_18\", \"FLAG_DOCUMENT_19\", \"FLAG_DOCUMENT_2\", \"FLAG_DOCUMENT_20\", \"FLAG_DOCUMENT_21\", \"FLAG_DOCUMENT_4\", \"FLAG_DOCUMENT_5\", \"FLAG_DOCUMENT_6\", \"FLAG_DOCUMENT_7\", \"FLAG_DOCUMENT_8\", \"FLAG_DOCUMENT_9\", \"FLAG_EMAIL\", \"FLAG_EMP_PHONE\", \"FLAG_MOBIL\", \"FLAG_OWN_CAR_N\", \"FLAG_OWN_REALTY_Y\", \"FLAG_WORK_PHONE\", \"LIVE_CITY_NOT_WORK_CITY\", \"LIVE_REGION_NOT_WORK_REGION\", \"NAME_EDUCATION_TYPE_Academic degree\", \"NAME_EDUCATION_TYPE_Lower secondary\", \"NAME_FAMILY_STATUS_Separated\", \"NAME_FAMILY_STATUS_Single / not married\", \"NAME_FAMILY_STATUS_Widow\", \"NAME_HOUSING_TYPE_Co-op apartment\", \"NAME_HOUSING_TYPE_House / apartment\", \"NAME_HOUSING_TYPE_Municipal apartment\", \"NAME_INCOME_TYPE_Businessman\", \"NAME_INCOME_TYPE_Commercial associate\", \"NAME_INCOME_TYPE_Pensioner\", \"NAME_INCOME_TYPE_State servant\", \"NAME_INCOME_TYPE_Student\", \"NAME_INCOME_TYPE_Unemployed\", \"NAME_TYPE_SUITE_Other_A\", \"NAME_TYPE_SUITE_Other_B\", \"NAME_TYPE_SUITE_Unaccompanied\", \"OCCUPATION_TYPE_Accountants\", \"OCCUPATION_TYPE_Cleaning staff\", \"OCCUPATION_TYPE_High skill tech staff\", \"OCCUPATION_TYPE_IT staff\", \"OCCUPATION_TYPE_Managers\", \"OCCUPATION_TYPE_Medicine staff\", \"OCCUPATION_TYPE_Private service staff\", \"OCCUPATION_TYPE_Realty agents\", \"OCCUPATION_TYPE_Secretaries\", \"ORGANIZATION_TYPE_Advertising\", \"ORGANIZATION_TYPE_Agriculture\", \"ORGANIZATION_TYPE_Business Entity Type 1\", \"ORGANIZATION_TYPE_Culture\", \"ORGANIZATION_TYPE_Emergency\", \"ORGANIZATION_TYPE_Government\", \"ORGANIZATION_TYPE_Hotel\", \"ORGANIZATION_TYPE_Industry: type 1\", \"ORGANIZATION_TYPE_Industry: type 10\", \"ORGANIZATION_TYPE_Industry: type 11\", \"ORGANIZATION_TYPE_Industry: type 12\", \"ORGANIZATION_TYPE_Industry: type 3\", \"ORGANIZATION_TYPE_Industry: type 4\", \"ORGANIZATION_TYPE_Industry: type 5\", \"ORGANIZATION_TYPE_Industry: type 6\", \"ORGANIZATION_TYPE_Industry: type 7\", \"ORGANIZATION_TYPE_Industry: type 8\", \"ORGANIZATION_TYPE_Industry: type 9\", \"ORGANIZATION_TYPE_Insurance\", \"ORGANIZATION_TYPE_Medicine\", \"ORGANIZATION_TYPE_Military\", \"ORGANIZATION_TYPE_Mobile\", \"ORGANIZATION_TYPE_Other\", \"ORGANIZATION_TYPE_Police\", \"ORGANIZATION_TYPE_Postal\", \"ORGANIZATION_TYPE_Realtor\", \"ORGANIZATION_TYPE_Religion\", \"ORGANIZATION_TYPE_School\", \"ORGANIZATION_TYPE_Security\", \"ORGANIZATION_TYPE_Security Ministries\", \"ORGANIZATION_TYPE_Services\", \"ORGANIZATION_TYPE_Telecom\", \"ORGANIZATION_TYPE_Trade: type 1\", \"ORGANIZATION_TYPE_Trade: type 2\", \"ORGANIZATION_TYPE_Trade: type 3\", \"ORGANIZATION_TYPE_Trade: type 4\", \"ORGANIZATION_TYPE_Trade: type 5\", \"ORGANIZATION_TYPE_Trade: type 6\", \"ORGANIZATION_TYPE_Trade: type 7\", \"ORGANIZATION_TYPE_Transport: type 1\", \"ORGANIZATION_TYPE_Transport: type 2\", \"ORGANIZATION_TYPE_Transport: type 4\", \"ORGANIZATION_TYPE_University\", \"ORGANIZATION_TYPE_XNA\", \"REGION_POPULATION_RELATIVE\", \"REGION_RATING_CLIENT\", \"REGION_RATING_CLIENT_W_CITY\", \"REG_CITY_NOT_LIVE_CITY\", \"REG_CITY_NOT_WORK_CITY\", \"REG_REGION_NOT_LIVE_REGION\", \"REG_REGION_NOT_WORK_REGION\", \"WEEKDAY_APPR_PROCESS_START_MONDAY\", \"WEEKDAY_APPR_PROCESS_START_SATURDAY\", \"WEEKDAY_APPR_PROCESS_START_SUNDAY\", \"WEEKDAY_APPR_PROCESS_START_THURSDAY\", \"YEAR_EMPLOYED\", \"NAME_CONTRACT_TYPE_Cash loans\", \"ORGANIZATION_TYPE_Bank\", \"ORGANIZATION_TYPE_Business Entity Type 3\", \"ORGANIZATION_TYPE_Restaurant\", \"AMT_CREDIT_LOG\", \"ORGANIZATION_TYPE_Housing\", \"ORGANIZATION_TYPE_Self-employed\", \"ORGANIZATION_TYPE_Construction\", \"OCCUPATION_TYPE_HR staff\", \"WEEKDAY_APPR_PROCESS_START_FRIDAY\", \"ORGANIZATION_TYPE_Business Entity Type 2\", \"AMT_INCOME_TOTAL_PER_MEMBER\", \"AMT_INCOME_TOTAL_POLY^3\", \"NAME_FAMILY_STATUS_Civil marriage\", \"WEEKDAY_APPR_PROCESS_START_WEDNESDAY\", \"NAME_TYPE_SUITE_Spouse, partner\", \"OCCUPATION_TYPE_Core staff\", \"NAME_EDUCATION_TYPE_Incomplete higher\", \"NAME_HOUSING_TYPE_Office apartment\", \"ORGANIZATION_TYPE_Electricity\", \"AMT_INCOME_TOTAL_POLY\", \"ORGANIZATION_TYPE_Industry: type 2\", \"OCCUPATION_TYPE_Security staff\", \"YEAR_REGISTRATION\", \"NAME_TYPE_SUITE_nan\", \"OCCUPATION_TYPE_Waiters/barmen staff\", \"WEEKDAY_APPR_PROCESS_START_TUESDAY\", \"NAME_CONTRACT_TYPE_Revolving loans\", \"OCCUPATION_TYPE_Sales staff\", \"OCCUPATION_TYPE_nan\", \"AMT_INCOME_TOTAL_LOG\", \"NAME_TYPE_SUITE_Group of people\", \"ORGANIZATION_TYPE_Cleaning\", \"OCCUPATION_TYPE_Drivers\", \"NAME_TYPE_SUITE_Children\", \"ORGANIZATION_TYPE_Kindergarten\", \"AMT_INCOME_TOTAL_POLY^2\", \"ORGANIZATION_TYPE_Legal Services\", \"AMT_ANNUITY_LOG\", \"NAME_INCOME_TYPE_Working\", \"ORGANIZATION_TYPE_Transport: type 3\", \"NAME_HOUSING_TYPE_Rented apartment\", \"FLAG_DOCUMENT_3\", \"OBS_30_CNT_SOCIAL_CIRCLE\", \"FLAG_OWN_REALTY_N\", \"YEARS_BEGINEXPLUATATION_AVG\", \"FLAG_PHONE\", \"NAME_EDUCATION_TYPE_Higher education\", \"OCCUPATION_TYPE_Cooking staff\", \"OCCUPATION_TYPE_Laborers\", \"OBS_60_CNT_SOCIAL_CIRCLE\", \"NAME_TYPE_SUITE_Family\", \"AMT_CREDIT\", \"FLAG_OWN_CAR_Y\", \"NAME_EDUCATION_TYPE_Secondary / secondary special\", \"NAME_HOUSING_TYPE_With parents\", \"NAME_FAMILY_STATUS_Married\", \"AMT_REQ_CREDIT_BUREAU_YEAR\", \"ORGANIZATION_TYPE_Industry: type 13\", \"CODE_GENDER_F\", \"AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY\", \"YEARS_BEGINEXPLUATATION_MODE\", \"AMT_INCOME_TOTAL\", \"AMT_CREDIT_POLY\", \"AMT_GOODS_PRICE\", \"DAYS_BIRTH\", \"AMT_ANNUITY\", \"HOUR_APPR_PROCESS_START\", \"CODE_GENDER_M\", \"OCCUPATION_TYPE_Low-skill Laborers\", \"AMT_CREDIT_POLY^3\", \"DAYS_LAST_PHONE_CHANGE\", \"AMT_CREDIT_POLY AMT_INCOME_TOTAL_POLY^2\", \"CREDIT_PER_MEMBER\", \"AMT_CREDIT_PER_AGE\", \"AMT_CREDIT_POLY^2\", \"AMT_INCOME_TOTAL_PER_AGE\", \"DAYS_REGISTRATION\", \"CREDIT_PER_INCOME\", \"YEARS_BEGINEXPLUATATION_MEDI\", \"AMT_CREDIT_PER_ANNUITY\", \"AMT_CREDIT_POLY^2 AMT_INCOME_TOTAL_POLY\", \"DAYS_EMPLOYED\", \"ANNUITY_PER_INCOME\", \"DAYS_ID_PUBLISH\", \"EXT_SOURCE_2\", \"EXT_SOURCE_3\"], \"type\": \"bar\", \"uid\": \"301d8e24-edeb-11e8-bfb1-605718f38acb\"}], {\"height\": 4000, \"margin\": {\"l\": 300}, \"title\": \"Import\\u00e2ncia das features\", \"width\": 900, \"yaxis\": {\"showgrid\": true}}, {\"showLink\": true, \"linkText\": \"Export to plot.ly\"})});</script>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "draw_feature_importance(gb.feature_importances_, x_train.columns)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Passo 1\n",
    "\n",
    "Encontrar o valor ideal para **n_estimators**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tempo de execução: 19.434 s\n",
      "Train Accuracy: 0.9237%\n",
      "Train Score Roc: 90.833%\n",
      "Train CV Score : Mean - 0.7184412 | Std - 0.01210847 | Min - 0.7064894 | Max - 0.7346434\n",
      "------------------------------------------------------------------------------------------------\n",
      "Test Accuracy: 0.919%\n",
      "Test Score Roc: 71.975%\n",
      "Test CV Score : Mean - 0.7060914 | Std - 0.03497484 | Min - 0.6561079 | Max - 0.7647783\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "[mean: 0.68768, std: 0.00839, params: {'n_estimators': 10},\n",
       " mean: 0.70333, std: 0.01258, params: {'n_estimators': 20},\n",
       " mean: 0.70740, std: 0.01106, params: {'n_estimators': 30},\n",
       " mean: 0.71668, std: 0.01087, params: {'n_estimators': 40},\n",
       " mean: 0.71789, std: 0.01181, params: {'n_estimators': 50},\n",
       " mean: 0.71792, std: 0.01264, params: {'n_estimators': 60},\n",
       " mean: 0.71656, std: 0.01174, params: {'n_estimators': 70},\n",
       " mean: 0.71995, std: 0.01038, params: {'n_estimators': 80},\n",
       " mean: 0.72076, std: 0.01042, params: {'n_estimators': 90}]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'n_estimators': 90}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "0.7207633871486057"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "t0 = time()\n",
    "param1 = {'n_estimators': np.arange(10, 100, 10)} \n",
    "grid1 = GridSearchCV(estimator = GradientBoostingClassifier(learning_rate=0.1, \n",
    "                                                            min_samples_split=100,\n",
    "                                                            min_samples_leaf=50,\n",
    "                                                            max_depth=5,\n",
    "                                                            max_features='sqrt',\n",
    "                                                            subsample=0.8,\n",
    "                                                            random_state=10), \n",
    "param_grid = param1, scoring='roc_auc', n_jobs=4, iid=False, cv=5)\n",
    "grid1.fit(x_train, y_train)\n",
    "print(\"Tempo de execução:\", round(time()-t0, 3), \"s\")\n",
    "draw_roc_score(grid1, x_train, y_train, x_test, y_test)\n",
    "display(grid1.grid_scores_, grid1.best_params_, grid1.best_score_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Valor ideal para n_estimators é 90."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Passo 2\n",
    "\n",
    "Encontrar os valores ideais para **max_depth** e **min_samples_split**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tempo de execução: 221.212 s\n",
      "Train Accuracy: 0.9202%\n",
      "Train Score Roc: 81.542%\n",
      "Train CV Score : Mean - 0.7263148 | Std - 0.01228269 | Min - 0.7122108 | Max - 0.7467664\n",
      "------------------------------------------------------------------------------------------------\n",
      "Test Accuracy: 0.9183333333333333%\n",
      "Test Score Roc: 73.084%\n",
      "Test CV Score : Mean - 0.7137096 | Std - 0.03478614 | Min - 0.6695605 | Max - 0.760093\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "[mean: 0.72336, std: 0.01489, params: {'max_depth': 2, 'min_samples_split': 100},\n",
       " mean: 0.72146, std: 0.01557, params: {'max_depth': 2, 'min_samples_split': 200},\n",
       " mean: 0.72218, std: 0.01550, params: {'max_depth': 2, 'min_samples_split': 300},\n",
       " mean: 0.72122, std: 0.01625, params: {'max_depth': 2, 'min_samples_split': 400},\n",
       " mean: 0.72170, std: 0.01654, params: {'max_depth': 2, 'min_samples_split': 500},\n",
       " mean: 0.72162, std: 0.01654, params: {'max_depth': 2, 'min_samples_split': 600},\n",
       " mean: 0.72101, std: 0.01544, params: {'max_depth': 2, 'min_samples_split': 700},\n",
       " mean: 0.72004, std: 0.01535, params: {'max_depth': 2, 'min_samples_split': 800},\n",
       " mean: 0.72164, std: 0.01591, params: {'max_depth': 2, 'min_samples_split': 900},\n",
       " mean: 0.72056, std: 0.01455, params: {'max_depth': 2, 'min_samples_split': 1000},\n",
       " mean: 0.72139, std: 0.01503, params: {'max_depth': 2, 'min_samples_split': 1100},\n",
       " mean: 0.72014, std: 0.01399, params: {'max_depth': 2, 'min_samples_split': 1200},\n",
       " mean: 0.72698, std: 0.01135, params: {'max_depth': 3, 'min_samples_split': 100},\n",
       " mean: 0.72879, std: 0.00968, params: {'max_depth': 3, 'min_samples_split': 200},\n",
       " mean: 0.72614, std: 0.01048, params: {'max_depth': 3, 'min_samples_split': 300},\n",
       " mean: 0.72504, std: 0.01046, params: {'max_depth': 3, 'min_samples_split': 400},\n",
       " mean: 0.72771, std: 0.01002, params: {'max_depth': 3, 'min_samples_split': 500},\n",
       " mean: 0.72951, std: 0.00968, params: {'max_depth': 3, 'min_samples_split': 600},\n",
       " mean: 0.72870, std: 0.01290, params: {'max_depth': 3, 'min_samples_split': 700},\n",
       " mean: 0.72797, std: 0.01104, params: {'max_depth': 3, 'min_samples_split': 800},\n",
       " mean: 0.73009, std: 0.00986, params: {'max_depth': 3, 'min_samples_split': 900},\n",
       " mean: 0.72526, std: 0.01473, params: {'max_depth': 3, 'min_samples_split': 1000},\n",
       " mean: 0.72604, std: 0.01015, params: {'max_depth': 3, 'min_samples_split': 1100},\n",
       " mean: 0.72641, std: 0.01003, params: {'max_depth': 3, 'min_samples_split': 1200},\n",
       " mean: 0.71562, std: 0.01592, params: {'max_depth': 4, 'min_samples_split': 100},\n",
       " mean: 0.72067, std: 0.01431, params: {'max_depth': 4, 'min_samples_split': 200},\n",
       " mean: 0.71726, std: 0.01528, params: {'max_depth': 4, 'min_samples_split': 300},\n",
       " mean: 0.71972, std: 0.01439, params: {'max_depth': 4, 'min_samples_split': 400},\n",
       " mean: 0.72312, std: 0.01078, params: {'max_depth': 4, 'min_samples_split': 500},\n",
       " mean: 0.71999, std: 0.01295, params: {'max_depth': 4, 'min_samples_split': 600},\n",
       " mean: 0.72262, std: 0.01021, params: {'max_depth': 4, 'min_samples_split': 700},\n",
       " mean: 0.72275, std: 0.01214, params: {'max_depth': 4, 'min_samples_split': 800},\n",
       " mean: 0.72042, std: 0.01243, params: {'max_depth': 4, 'min_samples_split': 900},\n",
       " mean: 0.72157, std: 0.01310, params: {'max_depth': 4, 'min_samples_split': 1000},\n",
       " mean: 0.71977, std: 0.01499, params: {'max_depth': 4, 'min_samples_split': 1100},\n",
       " mean: 0.72089, std: 0.01576, params: {'max_depth': 4, 'min_samples_split': 1200},\n",
       " mean: 0.71181, std: 0.01492, params: {'max_depth': 5, 'min_samples_split': 100},\n",
       " mean: 0.71773, std: 0.01703, params: {'max_depth': 5, 'min_samples_split': 200},\n",
       " mean: 0.71710, std: 0.01347, params: {'max_depth': 5, 'min_samples_split': 300},\n",
       " mean: 0.71927, std: 0.01584, params: {'max_depth': 5, 'min_samples_split': 400},\n",
       " mean: 0.71967, std: 0.01327, params: {'max_depth': 5, 'min_samples_split': 500},\n",
       " mean: 0.71780, std: 0.01422, params: {'max_depth': 5, 'min_samples_split': 600},\n",
       " mean: 0.72351, std: 0.00998, params: {'max_depth': 5, 'min_samples_split': 700},\n",
       " mean: 0.72476, std: 0.01310, params: {'max_depth': 5, 'min_samples_split': 800},\n",
       " mean: 0.72115, std: 0.01205, params: {'max_depth': 5, 'min_samples_split': 900},\n",
       " mean: 0.72251, std: 0.01695, params: {'max_depth': 5, 'min_samples_split': 1000},\n",
       " mean: 0.72339, std: 0.00967, params: {'max_depth': 5, 'min_samples_split': 1100},\n",
       " mean: 0.72375, std: 0.01386, params: {'max_depth': 5, 'min_samples_split': 1200},\n",
       " mean: 0.71882, std: 0.01213, params: {'max_depth': 6, 'min_samples_split': 100},\n",
       " mean: 0.71656, std: 0.01053, params: {'max_depth': 6, 'min_samples_split': 200},\n",
       " mean: 0.71994, std: 0.01064, params: {'max_depth': 6, 'min_samples_split': 300},\n",
       " mean: 0.71759, std: 0.01311, params: {'max_depth': 6, 'min_samples_split': 400},\n",
       " mean: 0.71555, std: 0.01320, params: {'max_depth': 6, 'min_samples_split': 500},\n",
       " mean: 0.72053, std: 0.00953, params: {'max_depth': 6, 'min_samples_split': 600},\n",
       " mean: 0.71856, std: 0.01194, params: {'max_depth': 6, 'min_samples_split': 700},\n",
       " mean: 0.72448, std: 0.01402, params: {'max_depth': 6, 'min_samples_split': 800},\n",
       " mean: 0.72053, std: 0.01241, params: {'max_depth': 6, 'min_samples_split': 900},\n",
       " mean: 0.72039, std: 0.01065, params: {'max_depth': 6, 'min_samples_split': 1000},\n",
       " mean: 0.72357, std: 0.01340, params: {'max_depth': 6, 'min_samples_split': 1100},\n",
       " mean: 0.72189, std: 0.01381, params: {'max_depth': 6, 'min_samples_split': 1200},\n",
       " mean: 0.70815, std: 0.01378, params: {'max_depth': 7, 'min_samples_split': 100},\n",
       " mean: 0.70679, std: 0.01382, params: {'max_depth': 7, 'min_samples_split': 200},\n",
       " mean: 0.71164, std: 0.01116, params: {'max_depth': 7, 'min_samples_split': 300},\n",
       " mean: 0.72000, std: 0.00716, params: {'max_depth': 7, 'min_samples_split': 400},\n",
       " mean: 0.71181, std: 0.01788, params: {'max_depth': 7, 'min_samples_split': 500},\n",
       " mean: 0.71546, std: 0.01577, params: {'max_depth': 7, 'min_samples_split': 600},\n",
       " mean: 0.71844, std: 0.01326, params: {'max_depth': 7, 'min_samples_split': 700},\n",
       " mean: 0.71951, std: 0.00887, params: {'max_depth': 7, 'min_samples_split': 800},\n",
       " mean: 0.72137, std: 0.01300, params: {'max_depth': 7, 'min_samples_split': 900},\n",
       " mean: 0.71452, std: 0.01150, params: {'max_depth': 7, 'min_samples_split': 1000},\n",
       " mean: 0.72213, std: 0.00912, params: {'max_depth': 7, 'min_samples_split': 1100},\n",
       " mean: 0.72138, std: 0.01801, params: {'max_depth': 7, 'min_samples_split': 1200},\n",
       " mean: 0.71086, std: 0.00798, params: {'max_depth': 8, 'min_samples_split': 100},\n",
       " mean: 0.70267, std: 0.01157, params: {'max_depth': 8, 'min_samples_split': 200},\n",
       " mean: 0.71007, std: 0.00946, params: {'max_depth': 8, 'min_samples_split': 300},\n",
       " mean: 0.70823, std: 0.01153, params: {'max_depth': 8, 'min_samples_split': 400},\n",
       " mean: 0.72003, std: 0.01200, params: {'max_depth': 8, 'min_samples_split': 500},\n",
       " mean: 0.71262, std: 0.01253, params: {'max_depth': 8, 'min_samples_split': 600},\n",
       " mean: 0.71679, std: 0.01439, params: {'max_depth': 8, 'min_samples_split': 700},\n",
       " mean: 0.71603, std: 0.00642, params: {'max_depth': 8, 'min_samples_split': 800},\n",
       " mean: 0.71861, std: 0.00657, params: {'max_depth': 8, 'min_samples_split': 900},\n",
       " mean: 0.71906, std: 0.01276, params: {'max_depth': 8, 'min_samples_split': 1000},\n",
       " mean: 0.71583, std: 0.01188, params: {'max_depth': 8, 'min_samples_split': 1100},\n",
       " mean: 0.72214, std: 0.01513, params: {'max_depth': 8, 'min_samples_split': 1200},\n",
       " mean: 0.70467, std: 0.01278, params: {'max_depth': 9, 'min_samples_split': 100},\n",
       " mean: 0.70260, std: 0.01600, params: {'max_depth': 9, 'min_samples_split': 200},\n",
       " mean: 0.70721, std: 0.01486, params: {'max_depth': 9, 'min_samples_split': 300},\n",
       " mean: 0.70494, std: 0.00974, params: {'max_depth': 9, 'min_samples_split': 400},\n",
       " mean: 0.70965, std: 0.01026, params: {'max_depth': 9, 'min_samples_split': 500},\n",
       " mean: 0.71482, std: 0.01359, params: {'max_depth': 9, 'min_samples_split': 600},\n",
       " mean: 0.71392, std: 0.00879, params: {'max_depth': 9, 'min_samples_split': 700},\n",
       " mean: 0.71583, std: 0.00893, params: {'max_depth': 9, 'min_samples_split': 800},\n",
       " mean: 0.72161, std: 0.01410, params: {'max_depth': 9, 'min_samples_split': 900},\n",
       " mean: 0.71397, std: 0.01367, params: {'max_depth': 9, 'min_samples_split': 1000},\n",
       " mean: 0.71670, std: 0.01229, params: {'max_depth': 9, 'min_samples_split': 1100},\n",
       " mean: 0.72064, std: 0.01238, params: {'max_depth': 9, 'min_samples_split': 1200}]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'max_depth': 3, 'min_samples_split': 900}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "0.7300935717269401"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "t0 = time()\n",
    "param2 = {'max_depth': np.arange(2, 10, 1), 'min_samples_split': np.arange(100, 1201, 100)} \n",
    "grid2 = GridSearchCV(estimator = GradientBoostingClassifier(learning_rate = 0.1,\n",
    "                                                                n_estimators = 90,\n",
    "                                                                max_features = 'sqrt',\n",
    "                                                                subsample = 0.8,\n",
    "                                                                random_state = 10), \n",
    "param_grid = param2, scoring = 'roc_auc', n_jobs = 4, iid = False, cv = 5) \n",
    "grid2.fit(x_train, y_train) \n",
    "print(\"Tempo de execução:\", round(time()-t0, 3), \"s\")\n",
    "draw_roc_score(grid2, x_train, y_train, x_test, y_test)\n",
    "display(grid2.grid_scores_, grid2.best_params_, grid2.best_score_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Valores ideais para max_depth é 3 e min_samples_split é 900"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Passo 3\n",
    "\n",
    "Encontrar o valor ideal para **min_samples_leaf**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tempo de execução: 32.375 s\n",
      "Train Accuracy: 0.9194%\n",
      "Train Score Roc: 81.276%\n",
      "Train CV Score : Mean - 0.7281602 | Std - 0.01097851 | Min - 0.7155317 | Max - 0.747107\n",
      "------------------------------------------------------------------------------------------------\n",
      "Test Accuracy: 0.9178333333333333%\n",
      "Test Score Roc: 73.032%\n",
      "Test CV Score : Mean - 0.7089171 | Std - 0.02979526 | Min - 0.6691021 | Max - 0.7553891\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "[mean: 0.72831, std: 0.00882, params: {'min_samples_leaf': 10},\n",
       " mean: 0.72870, std: 0.00944, params: {'min_samples_leaf': 15},\n",
       " mean: 0.72751, std: 0.01050, params: {'min_samples_leaf': 20},\n",
       " mean: 0.72735, std: 0.00974, params: {'min_samples_leaf': 25},\n",
       " mean: 0.72591, std: 0.00825, params: {'min_samples_leaf': 30},\n",
       " mean: 0.72957, std: 0.01081, params: {'min_samples_leaf': 35},\n",
       " mean: 0.72681, std: 0.01080, params: {'min_samples_leaf': 40},\n",
       " mean: 0.72928, std: 0.01053, params: {'min_samples_leaf': 45},\n",
       " mean: 0.72818, std: 0.01229, params: {'min_samples_leaf': 50},\n",
       " mean: 0.72750, std: 0.01381, params: {'min_samples_leaf': 55},\n",
       " mean: 0.72802, std: 0.01306, params: {'min_samples_leaf': 60},\n",
       " mean: 0.72842, std: 0.01288, params: {'min_samples_leaf': 65},\n",
       " mean: 0.72775, std: 0.01491, params: {'min_samples_leaf': 70},\n",
       " mean: 0.72829, std: 0.01260, params: {'min_samples_leaf': 75},\n",
       " mean: 0.72817, std: 0.01399, params: {'min_samples_leaf': 80},\n",
       " mean: 0.72839, std: 0.01357, params: {'min_samples_leaf': 85},\n",
       " mean: 0.72831, std: 0.01328, params: {'min_samples_leaf': 90},\n",
       " mean: 0.72717, std: 0.01217, params: {'min_samples_leaf': 95},\n",
       " mean: 0.72747, std: 0.01232, params: {'min_samples_leaf': 100}]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'min_samples_leaf': 35}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "0.7295729621436706"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "t0 = time()\n",
    "param3 = {'min_samples_leaf': np.arange(10,101, 5)} \n",
    "grid3 = GridSearchCV(estimator = GradientBoostingClassifier(learning_rate = 0.1,\n",
    "                                                                n_estimators = 90,\n",
    "                                                                max_depth=3,\n",
    "                                                                min_samples_split = 900,\n",
    "                                                                max_features = 'sqrt',\n",
    "                                                                subsample = 0.8,\n",
    "                                                                random_state = 10), \n",
    "param_grid = param3, scoring = 'roc_auc', n_jobs = 4, iid = False, cv = 5) \n",
    "grid3.fit(x_train, y_train) \n",
    "print(\"Tempo de execução:\", round(time()-t0, 3), \"s\")\n",
    "draw_roc_score(grid3, x_train, y_train, x_test, y_test)\n",
    "display(grid3.grid_scores_, grid3.best_params_, grid3.best_score_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "O valor ideal para min_samples_leaf foi 35, mas como a pontuação reduziu, irei rodar outro teste com um número elevado de min_samples_split, só que agora verificando o valor ideal junto com min_samples_leaf."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Passo 4\n",
    "\n",
    "Encontrar os valores ideais para **min_samples_leaf** e **min_samples_split**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tempo de execução: 171.18 s\n",
      "Train Accuracy: 0.9194%\n",
      "Train Score Roc: 80.58%\n",
      "Train CV Score : Mean - 0.7257094 | Std - 0.01311425 | Min - 0.7125849 | Max - 0.7436972\n",
      "------------------------------------------------------------------------------------------------\n",
      "Test Accuracy: 0.9185%\n",
      "Test Score Roc: 73.421%\n",
      "Test CV Score : Mean - 0.7177262 | Std - 0.02977896 | Min - 0.6792804 | Max - 0.7624634\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "[mean: 0.72926, std: 0.01137, params: {'min_samples_leaf': 10, 'min_samples_split': 1000},\n",
       " mean: 0.72977, std: 0.01007, params: {'min_samples_leaf': 10, 'min_samples_split': 1100},\n",
       " mean: 0.72964, std: 0.01034, params: {'min_samples_leaf': 10, 'min_samples_split': 1200},\n",
       " mean: 0.72897, std: 0.01168, params: {'min_samples_leaf': 10, 'min_samples_split': 1300},\n",
       " mean: 0.72762, std: 0.01250, params: {'min_samples_leaf': 10, 'min_samples_split': 1400},\n",
       " mean: 0.72941, std: 0.01251, params: {'min_samples_leaf': 10, 'min_samples_split': 1500},\n",
       " mean: 0.73122, std: 0.01234, params: {'min_samples_leaf': 10, 'min_samples_split': 1600},\n",
       " mean: 0.73171, std: 0.01492, params: {'min_samples_leaf': 10, 'min_samples_split': 1700},\n",
       " mean: 0.73013, std: 0.01372, params: {'min_samples_leaf': 10, 'min_samples_split': 1800},\n",
       " mean: 0.72937, std: 0.01215, params: {'min_samples_leaf': 10, 'min_samples_split': 1900},\n",
       " mean: 0.72922, std: 0.01327, params: {'min_samples_leaf': 10, 'min_samples_split': 2000},\n",
       " mean: 0.72914, std: 0.01306, params: {'min_samples_leaf': 10, 'min_samples_split': 2100},\n",
       " mean: 0.72589, std: 0.01238, params: {'min_samples_leaf': 10, 'min_samples_split': 2200},\n",
       " mean: 0.72750, std: 0.00995, params: {'min_samples_leaf': 20, 'min_samples_split': 1000},\n",
       " mean: 0.73098, std: 0.01026, params: {'min_samples_leaf': 20, 'min_samples_split': 1100},\n",
       " mean: 0.72855, std: 0.01275, params: {'min_samples_leaf': 20, 'min_samples_split': 1200},\n",
       " mean: 0.73068, std: 0.01173, params: {'min_samples_leaf': 20, 'min_samples_split': 1300},\n",
       " mean: 0.72850, std: 0.01406, params: {'min_samples_leaf': 20, 'min_samples_split': 1400},\n",
       " mean: 0.73074, std: 0.01212, params: {'min_samples_leaf': 20, 'min_samples_split': 1500},\n",
       " mean: 0.72959, std: 0.01404, params: {'min_samples_leaf': 20, 'min_samples_split': 1600},\n",
       " mean: 0.72926, std: 0.01400, params: {'min_samples_leaf': 20, 'min_samples_split': 1700},\n",
       " mean: 0.73057, std: 0.01488, params: {'min_samples_leaf': 20, 'min_samples_split': 1800},\n",
       " mean: 0.72943, std: 0.01106, params: {'min_samples_leaf': 20, 'min_samples_split': 1900},\n",
       " mean: 0.73063, std: 0.01315, params: {'min_samples_leaf': 20, 'min_samples_split': 2000},\n",
       " mean: 0.72908, std: 0.01068, params: {'min_samples_leaf': 20, 'min_samples_split': 2100},\n",
       " mean: 0.72687, std: 0.01048, params: {'min_samples_leaf': 20, 'min_samples_split': 2200},\n",
       " mean: 0.72664, std: 0.00938, params: {'min_samples_leaf': 30, 'min_samples_split': 1000},\n",
       " mean: 0.72864, std: 0.01026, params: {'min_samples_leaf': 30, 'min_samples_split': 1100},\n",
       " mean: 0.72642, std: 0.01211, params: {'min_samples_leaf': 30, 'min_samples_split': 1200},\n",
       " mean: 0.72931, std: 0.01055, params: {'min_samples_leaf': 30, 'min_samples_split': 1300},\n",
       " mean: 0.72831, std: 0.01286, params: {'min_samples_leaf': 30, 'min_samples_split': 1400},\n",
       " mean: 0.72965, std: 0.01257, params: {'min_samples_leaf': 30, 'min_samples_split': 1500},\n",
       " mean: 0.72874, std: 0.01488, params: {'min_samples_leaf': 30, 'min_samples_split': 1600},\n",
       " mean: 0.72894, std: 0.01518, params: {'min_samples_leaf': 30, 'min_samples_split': 1700},\n",
       " mean: 0.72844, std: 0.01369, params: {'min_samples_leaf': 30, 'min_samples_split': 1800},\n",
       " mean: 0.72879, std: 0.01270, params: {'min_samples_leaf': 30, 'min_samples_split': 1900},\n",
       " mean: 0.72915, std: 0.01250, params: {'min_samples_leaf': 30, 'min_samples_split': 2000},\n",
       " mean: 0.72688, std: 0.01146, params: {'min_samples_leaf': 30, 'min_samples_split': 2100},\n",
       " mean: 0.72554, std: 0.01332, params: {'min_samples_leaf': 30, 'min_samples_split': 2200},\n",
       " mean: 0.72837, std: 0.01157, params: {'min_samples_leaf': 40, 'min_samples_split': 1000},\n",
       " mean: 0.72861, std: 0.01142, params: {'min_samples_leaf': 40, 'min_samples_split': 1100},\n",
       " mean: 0.72904, std: 0.00925, params: {'min_samples_leaf': 40, 'min_samples_split': 1200},\n",
       " mean: 0.72799, std: 0.01200, params: {'min_samples_leaf': 40, 'min_samples_split': 1300},\n",
       " mean: 0.72869, std: 0.01142, params: {'min_samples_leaf': 40, 'min_samples_split': 1400},\n",
       " mean: 0.73076, std: 0.01227, params: {'min_samples_leaf': 40, 'min_samples_split': 1500},\n",
       " mean: 0.72769, std: 0.01454, params: {'min_samples_leaf': 40, 'min_samples_split': 1600},\n",
       " mean: 0.72986, std: 0.01422, params: {'min_samples_leaf': 40, 'min_samples_split': 1700},\n",
       " mean: 0.73083, std: 0.01329, params: {'min_samples_leaf': 40, 'min_samples_split': 1800},\n",
       " mean: 0.73048, std: 0.01339, params: {'min_samples_leaf': 40, 'min_samples_split': 1900},\n",
       " mean: 0.72786, std: 0.01204, params: {'min_samples_leaf': 40, 'min_samples_split': 2000},\n",
       " mean: 0.72778, std: 0.01106, params: {'min_samples_leaf': 40, 'min_samples_split': 2100},\n",
       " mean: 0.72641, std: 0.01237, params: {'min_samples_leaf': 40, 'min_samples_split': 2200},\n",
       " mean: 0.72826, std: 0.01265, params: {'min_samples_leaf': 50, 'min_samples_split': 1000},\n",
       " mean: 0.72954, std: 0.01128, params: {'min_samples_leaf': 50, 'min_samples_split': 1100},\n",
       " mean: 0.72805, std: 0.01388, params: {'min_samples_leaf': 50, 'min_samples_split': 1200},\n",
       " mean: 0.72880, std: 0.01365, params: {'min_samples_leaf': 50, 'min_samples_split': 1300},\n",
       " mean: 0.72955, std: 0.01482, params: {'min_samples_leaf': 50, 'min_samples_split': 1400},\n",
       " mean: 0.72937, std: 0.01373, params: {'min_samples_leaf': 50, 'min_samples_split': 1500},\n",
       " mean: 0.72746, std: 0.01427, params: {'min_samples_leaf': 50, 'min_samples_split': 1600},\n",
       " mean: 0.72923, std: 0.01466, params: {'min_samples_leaf': 50, 'min_samples_split': 1700},\n",
       " mean: 0.72969, std: 0.01522, params: {'min_samples_leaf': 50, 'min_samples_split': 1800},\n",
       " mean: 0.72781, std: 0.01449, params: {'min_samples_leaf': 50, 'min_samples_split': 1900},\n",
       " mean: 0.72815, std: 0.01306, params: {'min_samples_leaf': 50, 'min_samples_split': 2000},\n",
       " mean: 0.72595, std: 0.01267, params: {'min_samples_leaf': 50, 'min_samples_split': 2100},\n",
       " mean: 0.72626, std: 0.01369, params: {'min_samples_leaf': 50, 'min_samples_split': 2200},\n",
       " mean: 0.72906, std: 0.01377, params: {'min_samples_leaf': 60, 'min_samples_split': 1000},\n",
       " mean: 0.72895, std: 0.01078, params: {'min_samples_leaf': 60, 'min_samples_split': 1100},\n",
       " mean: 0.73161, std: 0.01185, params: {'min_samples_leaf': 60, 'min_samples_split': 1200},\n",
       " mean: 0.72996, std: 0.01286, params: {'min_samples_leaf': 60, 'min_samples_split': 1300},\n",
       " mean: 0.73089, std: 0.01245, params: {'min_samples_leaf': 60, 'min_samples_split': 1400},\n",
       " mean: 0.72837, std: 0.01423, params: {'min_samples_leaf': 60, 'min_samples_split': 1500},\n",
       " mean: 0.73068, std: 0.01363, params: {'min_samples_leaf': 60, 'min_samples_split': 1600},\n",
       " mean: 0.73060, std: 0.01433, params: {'min_samples_leaf': 60, 'min_samples_split': 1700},\n",
       " mean: 0.72929, std: 0.01560, params: {'min_samples_leaf': 60, 'min_samples_split': 1800},\n",
       " mean: 0.72833, std: 0.01312, params: {'min_samples_leaf': 60, 'min_samples_split': 1900},\n",
       " mean: 0.72828, std: 0.01254, params: {'min_samples_leaf': 60, 'min_samples_split': 2000},\n",
       " mean: 0.72716, std: 0.01381, params: {'min_samples_leaf': 60, 'min_samples_split': 2100},\n",
       " mean: 0.72672, std: 0.01367, params: {'min_samples_leaf': 60, 'min_samples_split': 2200},\n",
       " mean: 0.72793, std: 0.01477, params: {'min_samples_leaf': 70, 'min_samples_split': 1000},\n",
       " mean: 0.72875, std: 0.01271, params: {'min_samples_leaf': 70, 'min_samples_split': 1100},\n",
       " mean: 0.72960, std: 0.01441, params: {'min_samples_leaf': 70, 'min_samples_split': 1200},\n",
       " mean: 0.72971, std: 0.01398, params: {'min_samples_leaf': 70, 'min_samples_split': 1300},\n",
       " mean: 0.72968, std: 0.01406, params: {'min_samples_leaf': 70, 'min_samples_split': 1400},\n",
       " mean: 0.72835, std: 0.01450, params: {'min_samples_leaf': 70, 'min_samples_split': 1500},\n",
       " mean: 0.73020, std: 0.01534, params: {'min_samples_leaf': 70, 'min_samples_split': 1600},\n",
       " mean: 0.73019, std: 0.01413, params: {'min_samples_leaf': 70, 'min_samples_split': 1700},\n",
       " mean: 0.73121, std: 0.01433, params: {'min_samples_leaf': 70, 'min_samples_split': 1800},\n",
       " mean: 0.72847, std: 0.01315, params: {'min_samples_leaf': 70, 'min_samples_split': 1900},\n",
       " mean: 0.72901, std: 0.01255, params: {'min_samples_leaf': 70, 'min_samples_split': 2000},\n",
       " mean: 0.72725, std: 0.01332, params: {'min_samples_leaf': 70, 'min_samples_split': 2100},\n",
       " mean: 0.72549, std: 0.01379, params: {'min_samples_leaf': 70, 'min_samples_split': 2200},\n",
       " mean: 0.72684, std: 0.01376, params: {'min_samples_leaf': 80, 'min_samples_split': 1000},\n",
       " mean: 0.72890, std: 0.01131, params: {'min_samples_leaf': 80, 'min_samples_split': 1100},\n",
       " mean: 0.72714, std: 0.01494, params: {'min_samples_leaf': 80, 'min_samples_split': 1200},\n",
       " mean: 0.72823, std: 0.01412, params: {'min_samples_leaf': 80, 'min_samples_split': 1300},\n",
       " mean: 0.72692, std: 0.01418, params: {'min_samples_leaf': 80, 'min_samples_split': 1400},\n",
       " mean: 0.72801, std: 0.01591, params: {'min_samples_leaf': 80, 'min_samples_split': 1500},\n",
       " mean: 0.73056, std: 0.01371, params: {'min_samples_leaf': 80, 'min_samples_split': 1600},\n",
       " mean: 0.73006, std: 0.01443, params: {'min_samples_leaf': 80, 'min_samples_split': 1700},\n",
       " mean: 0.72950, std: 0.01330, params: {'min_samples_leaf': 80, 'min_samples_split': 1800},\n",
       " mean: 0.72836, std: 0.01244, params: {'min_samples_leaf': 80, 'min_samples_split': 1900},\n",
       " mean: 0.72846, std: 0.01310, params: {'min_samples_leaf': 80, 'min_samples_split': 2000},\n",
       " mean: 0.72660, std: 0.01309, params: {'min_samples_leaf': 80, 'min_samples_split': 2100},\n",
       " mean: 0.72466, std: 0.01472, params: {'min_samples_leaf': 80, 'min_samples_split': 2200},\n",
       " mean: 0.72729, std: 0.01574, params: {'min_samples_leaf': 90, 'min_samples_split': 1000},\n",
       " mean: 0.72959, std: 0.01179, params: {'min_samples_leaf': 90, 'min_samples_split': 1100},\n",
       " mean: 0.72784, std: 0.01398, params: {'min_samples_leaf': 90, 'min_samples_split': 1200},\n",
       " mean: 0.72703, std: 0.01442, params: {'min_samples_leaf': 90, 'min_samples_split': 1300},\n",
       " mean: 0.72664, std: 0.01538, params: {'min_samples_leaf': 90, 'min_samples_split': 1400},\n",
       " mean: 0.72925, std: 0.01648, params: {'min_samples_leaf': 90, 'min_samples_split': 1500},\n",
       " mean: 0.73087, std: 0.01426, params: {'min_samples_leaf': 90, 'min_samples_split': 1600},\n",
       " mean: 0.73045, std: 0.01483, params: {'min_samples_leaf': 90, 'min_samples_split': 1700},\n",
       " mean: 0.73081, std: 0.01392, params: {'min_samples_leaf': 90, 'min_samples_split': 1800},\n",
       " mean: 0.72968, std: 0.01401, params: {'min_samples_leaf': 90, 'min_samples_split': 1900},\n",
       " mean: 0.72814, std: 0.01338, params: {'min_samples_leaf': 90, 'min_samples_split': 2000},\n",
       " mean: 0.72670, std: 0.01254, params: {'min_samples_leaf': 90, 'min_samples_split': 2100},\n",
       " mean: 0.72510, std: 0.01366, params: {'min_samples_leaf': 90, 'min_samples_split': 2200},\n",
       " mean: 0.72822, std: 0.01489, params: {'min_samples_leaf': 100, 'min_samples_split': 1000},\n",
       " mean: 0.72711, std: 0.01245, params: {'min_samples_leaf': 100, 'min_samples_split': 1100},\n",
       " mean: 0.72635, std: 0.01550, params: {'min_samples_leaf': 100, 'min_samples_split': 1200},\n",
       " mean: 0.72838, std: 0.01349, params: {'min_samples_leaf': 100, 'min_samples_split': 1300},\n",
       " mean: 0.72856, std: 0.01479, params: {'min_samples_leaf': 100, 'min_samples_split': 1400},\n",
       " mean: 0.72916, std: 0.01534, params: {'min_samples_leaf': 100, 'min_samples_split': 1500},\n",
       " mean: 0.72994, std: 0.01698, params: {'min_samples_leaf': 100, 'min_samples_split': 1600},\n",
       " mean: 0.72958, std: 0.01520, params: {'min_samples_leaf': 100, 'min_samples_split': 1700},\n",
       " mean: 0.73100, std: 0.01614, params: {'min_samples_leaf': 100, 'min_samples_split': 1800},\n",
       " mean: 0.72888, std: 0.01349, params: {'min_samples_leaf': 100, 'min_samples_split': 1900},\n",
       " mean: 0.72841, std: 0.01386, params: {'min_samples_leaf': 100, 'min_samples_split': 2000},\n",
       " mean: 0.72735, std: 0.01247, params: {'min_samples_leaf': 100, 'min_samples_split': 2100},\n",
       " mean: 0.72630, std: 0.01354, params: {'min_samples_leaf': 100, 'min_samples_split': 2200}]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'min_samples_leaf': 10, 'min_samples_split': 1700}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "0.7317122629931205"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "t0 = time()\n",
    "param4 = {'min_samples_leaf': np.arange(10,101, 10), 'min_samples_split': np.arange(1000, 2201, 100)} \n",
    "grid4 = GridSearchCV(estimator = GradientBoostingClassifier(learning_rate = 0.1,\n",
    "                                                                n_estimators = 90,\n",
    "                                                                max_depth=3,\n",
    "                                                                max_features = 'sqrt',\n",
    "                                                                subsample = 0.8,\n",
    "                                                                random_state = 10), \n",
    "param_grid = param4, scoring = 'roc_auc', n_jobs = 4, iid = False, cv = 5) \n",
    "grid4.fit(x_train, y_train) \n",
    "print(\"Tempo de execução:\", round(time()-t0, 3), \"s\")\n",
    "draw_roc_score(grid4, x_train, y_train, x_test, y_test)\n",
    "display(grid4.grid_scores_, grid4.best_params_, grid4.best_score_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "O score voltou a subir e o valor ideal para min_samples_leaf ficou 10 e min_samples_split com 1700"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Passo 5\n",
    "\n",
    "Encontrar o valor ideaL para **max_features**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tempo de execução: 53.231 s\n",
      "Train Accuracy: 0.9194%\n",
      "Train Score Roc: 80.567%\n",
      "Train CV Score : Mean - 0.7272665 | Std - 0.01277376 | Min - 0.7132762 | Max - 0.7492578\n",
      "------------------------------------------------------------------------------------------------\n",
      "Test Accuracy: 0.918%\n",
      "Test Score Roc: 74.005%\n",
      "Test CV Score : Mean - 0.7233044 | Std - 0.03284488 | Min - 0.6746029 | Max - 0.7600144\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "[mean: 0.67519, std: 0.00528, params: {'max_features': 1},\n",
       " mean: 0.68995, std: 0.00690, params: {'max_features': 2},\n",
       " mean: 0.69069, std: 0.00811, params: {'max_features': 3},\n",
       " mean: 0.70191, std: 0.01388, params: {'max_features': 4},\n",
       " mean: 0.70131, std: 0.01450, params: {'max_features': 5},\n",
       " mean: 0.70787, std: 0.00846, params: {'max_features': 6},\n",
       " mean: 0.71294, std: 0.01542, params: {'max_features': 7},\n",
       " mean: 0.71862, std: 0.01014, params: {'max_features': 8},\n",
       " mean: 0.71660, std: 0.01129, params: {'max_features': 9},\n",
       " mean: 0.72084, std: 0.01110, params: {'max_features': 10},\n",
       " mean: 0.71493, std: 0.01232, params: {'max_features': 11},\n",
       " mean: 0.71993, std: 0.01045, params: {'max_features': 12},\n",
       " mean: 0.72075, std: 0.00988, params: {'max_features': 13},\n",
       " mean: 0.71874, std: 0.01173, params: {'max_features': 14},\n",
       " mean: 0.72607, std: 0.01277, params: {'max_features': 15},\n",
       " mean: 0.73171, std: 0.01492, params: {'max_features': 16},\n",
       " mean: 0.72787, std: 0.01389, params: {'max_features': 17},\n",
       " mean: 0.72932, std: 0.01084, params: {'max_features': 18},\n",
       " mean: 0.73112, std: 0.00969, params: {'max_features': 19},\n",
       " mean: 0.72601, std: 0.00997, params: {'max_features': 20},\n",
       " mean: 0.72965, std: 0.01482, params: {'max_features': 21},\n",
       " mean: 0.72757, std: 0.01355, params: {'max_features': 22},\n",
       " mean: 0.72695, std: 0.01107, params: {'max_features': 23},\n",
       " mean: 0.73225, std: 0.01050, params: {'max_features': 24},\n",
       " mean: 0.72518, std: 0.01092, params: {'max_features': 25},\n",
       " mean: 0.72671, std: 0.01176, params: {'max_features': 26},\n",
       " mean: 0.72680, std: 0.01281, params: {'max_features': 27},\n",
       " mean: 0.72911, std: 0.00998, params: {'max_features': 28},\n",
       " mean: 0.73053, std: 0.01465, params: {'max_features': 29}]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'max_features': 24}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "0.7322463965050072"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "t0 = time()\n",
    "param5 = {'max_features': np.arange(1, 30, 1)} \n",
    "grid5 = GridSearchCV(estimator = GradientBoostingClassifier(learning_rate = 0.1,\n",
    "                                                                n_estimators = 90,\n",
    "                                                                max_depth=3,\n",
    "                                                                max_features = 'sqrt',\n",
    "                                                                subsample = 0.8,\n",
    "                                                                min_samples_leaf= 10,\n",
    "                                                                min_samples_split = 1700,\n",
    "                                                                random_state = 10), \n",
    "param_grid = param5, scoring = 'roc_auc', n_jobs = 4, iid = False, cv = 5) \n",
    "grid5.fit(x_train, y_train) \n",
    "print(\"Tempo de execução:\", round(time()-t0, 3), \"s\")\n",
    "draw_roc_score(grid5, x_train, y_train, x_test, y_test)\n",
    "display(grid5.grid_scores_, grid5.best_params_, grid5.best_score_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "O valor ideal para max_features é 24"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Passo 6\n",
    "\n",
    "Encontrar o valor ideal para **learning_rate**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tempo de execução: 17.521 s\n",
      "Train Accuracy: 0.9197%\n",
      "Train Score Roc: 80.105%\n",
      "Train CV Score : Mean - 0.7298058 | Std - 0.01249685 | Min - 0.7179584 | Max - 0.7480388\n",
      "------------------------------------------------------------------------------------------------\n",
      "Test Accuracy: 0.9183333333333333%\n",
      "Test Score Roc: 73.892%\n",
      "Test CV Score : Mean - 0.7218239 | Std - 0.03328006 | Min - 0.6703837 | Max - 0.7578682\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYoAAAEWCAYAAAB42tAoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzs3Xd4FOX2wPHvIYChCQqC99LkCiogihBpAgFBpCioIE3pAQEVGxYs194VvSpXQIqI+EMuiAKCcEW69N6RciWgSJEuhCSc3x8zIUtINpuQrTmf58kzO7OzsydD2LPzvvOeV1QVY4wxJiN5gh2AMcaY0GaJwhhjjFeWKIwxxnhlicIYY4xXliiMMcZ4ZYnCGGOMV5YojAlhIjJLRO4Ldhwmd7NEkUuJSH0R+VlEjorInyKySERuDnJM3UUkWUROiMgxEVkrIncEM6asEJGhbuwnROSMiCR6rM/IzjFVtZmqjsvpWH0lIi94/A6nPf59TojI2os47h0isiUnYzX+Y4kiFxKRS4FpwMfA5UBp4GUgIYffJyobL1usqoWBYsC/gfEiUiwn4/IXVe2rqoXd+N8Avk5ZV9UWafcXkbyBjzJrVPVVj9/pIWCBx+90Y7DjM4FhiSJ3ugZAVf9PVZNV9ZSqzlLVdSk7iEhvEdksIsdFZJOI1HC3VxaRuSJyREQ2ikhrj9d8LiKfish0ETkJNBaRS0TkPRHZLSJ/uN+6C2QWoKqeBcYChYBKHu/R2n3fI24clT2eKysi34jIARE5JCKfpHdsEckjIs+IyA53vwkicrn73FUioiLSQ0TiReSwiPQVkZtFZJ37vukeNzMiUtHj2LuBWe72W0RkiXvsNSLS0OM1C0Wku/s4TkTmicgH7r47RaSZx75lRGSae4X4i4j09BJLtIgMdn/HP0Tk3yISnc3f6wYRmeOeq00icqfHc/eIyFb37yheRB4UkZLAf4BrPK5OiopIXhF5SUR2ichBERnrfqkxwaaq9pPLfoBLgUPAGKAFcFma5+8F9gI3AwJUBMoD+YDtwLNAfuBW4Dhwrfu6z4GjwC04X0KigQ+BKThXLkWAqcCbGcTVHVjoPo4CHgTOACXdbdcAJ4Hb3FiecuPJ7+6/FvgAJ7lEA/UzeJ9HgSVAGeASYBjwf+5zVwEKDHWP0Qw4DXwLlMS5+toPxGZyjl8CvkyzraJ77NFAQaAAUNb9t7jdPWfNgYNAcfc1C4Hu7uM4IBHo6f6+DwPxHsdfhHOVGA3UcI+TbpzAJ8Bk4DL372E68Gomv1McMDfNtmLAPqCjG1Md4E+ggvu3cwSo6e5bAqjuPr4D2JLmWM8Dc4Er3XMzFvgs2P9f7EctUeTWH6Cy+8G+B0hyP8xLuc/NBB5J5zUN3A+FPB7b/g94yX38OfCFx3PifrBf7bGtLrArg5i6u7EccT8QTwHtPZ5/AZjgsZ4HJ6E1co97AMjrw+++GWjisf439/3ykpooSns8fwjo4LE+CXg0k/d4iYwTRTmPbc8Bo9PsNxu4z32cNlFs8djvUvd4JdwP5kSgkMfz7wIj0oktD07yK5/m3/aXTH6n9BJFL2BGmm3jgCfcf/+DQFegcJp90ksU8UBtj/VKwPFg/1+xHyXk20iNf6jqZpwPZkTkOuBLnG//nXC+5e5I52V/x/kGe9Zj268437JTxHs8vgLnm/NKEUnZJjjfPDOyRFXri0hhYCTOB9gEj/f/1eN3OCsi8e77JwK/qmqSl2OnKA9MFhHP3yMZKOWx/ofH41PprBf24X0y4nmOygOdRORuj235gB8yeO0+j8d/ucvCOOfmoKqe9Hj+V+D6dI5xJc6V1No0/y7ZUR64VUSOeGzLC/yuqioibXCuQP8lIquAJ1V1VdqDuP1ZpYFZIuJZqTSfiBRT1SNpX2MCxxKFQVW3iMjnwAPupnjg6nR2/Q0oKyJ5PJJFOWCb5+E8Hh/E+VCtqqp7sxjTCRHpD+wQkVGqutp9/2op+4jzKVcW56oiASgnInl9SBbxQE9VXZT2CRG5KitxZoe6X5c9Yhmtqv0u8rC/ASVEpJBHsiiHc27S+gOnSe9aVf0jneezIh6Yrqp3p/eke45biUh+4Emcq43KnP93gqomi8jvQEtVzfbdVMY/rDM7FxKR60TkCREp466XxbmSWOLuMgIYKCI1xVFRRMoDS3Gakp4SkXwi0gi4Exif3vu4yeQz4AO3AxMRKS0it/sSp6oecmP5p7tpAs6HThMRyYfTvJEA/AwsA34H3hKRQm5n7S0ZHHoo8Lr7OyEiV7jffINhLHC3iNwmIlFu3I1F5O9ZOYiq7gJWAG+IcwNBdaAHzgdz2n2Tcc7rh+7vLm5HeLO0+/pgElBLRNq5ndH5RaSu+zdTRETai0gRnCu+EzhXbuAkq1IiUsjjWEOBt0WkNICIlJIwuj06klmiyJ2OA7WBpeLcnbQE2IDzwYuq/gd4HfjK3fdb4HJVPQO0xukAP4hz+2pXVfV2P/zTOB3OS0TkGPAjcG0WYv0QaCkiN6jqVuB+nA7bgzhJ6k5VPeN++N2J0w+wG6fvpUMGx/wXTp/MLBE57v7+tbMQU45R1f8Bd+P0vxzAif0Jsvd/swNOu/4+YCLwrKrOyWDfJ3Cappbh3IAwC4+7y3ylqn/idMT3dt/3N+AVnOYzgD44Vx1HcDq8e7jbV+I0r+127+AqCrwJzAfmuX8rC4GbshqTyXly/lWwMcYYcz67ojDGGOOV3xKFiIwSkf0isiGD50VEPhKR7eIMZKrhr1iMMcZknz+vKD7HGTyUkRY4baKVcNoxP/VjLMYYY7LJb4lCVefjjNDMSBucwVmqqkuAYiLyN3/FY4wxJnuCOY6iNOcPPNrjbvs97Y4i0gfnqoNChQrVvO666wISoDHGZCQ5OfN9PJ0+DYmJqY8TEkDSDHNMSIBjx7IXTwkOcHk6383/5ARHgCRnQOYV2Tl2MBNFeiNB070FS1WHA8MBYmJidMWKFf6MyxgTgY4dg5UrL9y+bh388Ufqh/aUKRAdfeGHuKfly3MmppIlz1/P634it24Nd92V/msqzRnOjRu/In/+87dfssT55RLqxKaUQEFEGL7vN/ZWvJa3f5z2a9pj+SqYiWIPzqjaFGVw7sE2xpiLdvYszJoF06bBkCGZ75/yIZ3kjutvcUFh+FTNm0OBAlC/ftbiqVgR/vEPZ/3KK9MkiuHD4auvnMdHcUp2pmfePGcZG3v+9thY6NyZg61a0a9fPzp06MB9993Hw+7Tb3vLfJkIZqKYAjwkIuNxBjsdVdULmp2MMUYV9uxxrgjGjHE+pL2ZNAnOnDl/W9260KpV+h/u11wDf/NnD6lnEshIRgkgLTch0KfPeZtVlREjRjCwShUSExNp1arVRQR8Pr8lChH5P5yqniVEZA/wIu5oTVUdilPWuCXOqN2/SB2xaYyJQImJ0Lev06zz++/O5+Lll0MeH26p2bbtwm2VvIwjL1vWeb9bb4V+/eDGG+GSS7Ife5aklxR8SQIZJABf7Nixg969ezNnzhwaN27MZ599xtVXp1euLXv8lihUtVMmzyvOfAMXLTExkT179nD69OmcOFxYi46OpkyZMuTLly/znY3xs+3bYe9epx9gwIDU7cWKwdGjUKUKlCmT+XFq1HA6elu0gJo14aabvPchBIyvSeEikoAv1q9fz8qVKxk+fDhxcXFIDp+csCvhkV5n9q5duyhSpAjFixfP8RMUTlSVQ4cOcfz4cSpUqBDscEwusG0bfPst7NjhPE5pEpqRwQzhpUtDfHyIfMhnR9rEkNGVgh+TQooNGzawatUqunbtCsChQ4coXrx4hvuLyEpVjcnOe0VEmfHTp09z1VVX5eokAc4dDsWLF+fAgQPBDsVEmDNnYP9+WL0aliyBrVth/foLm4RuuMFp4rnpJihaFGJinG1lyjidt5Urp3/8kJFZX0LaxODnK4X0nDlzhjfeeIM33niDUqVK0b59e6Kjo70miYsVEYkCyPVJIoWdB3OxkpJg0CDnXv8FC+D4cdi588L9ypVzksKrr8IDDzh9D2lv2QwLnskhs76EICQGT0uXLqVXr15s3LiR+++/nw8++IDo6GxNdZ4lEZMojDFZt38/vPMOFCzo3Fn05Zfwv/+lPn/ppc74g/btnURQrx5Urw7XXedcMYQlb81HQU4E3uzdu5cGDRpQqlQppk2blqN3NWXGEkUOOHToEE2aNAFg3759REVFccUVzgDIZcuWkd+Hr1k9evTgmWee4dprszJVgzFZd/QofPYZPPlk+s8XKeI0GX3/fea3oYakMGg+yopt27ZxzTXXULp0ab7++muaNGnCpZdeGtAYLFHkgOLFi7NmzRoAXnrpJQoXLszAgQPP2ydlkvI8GdwLOHr0aL/HaXK3o0fhoYecq4YUUVEweDA8/HAYdzCnSEkQId585KsjR47w1FNPMWLECObOnUvDhg25++50Z5z1O0sUfrR9+3buuusu6tevz9KlS5k2bRrr1q3jlVdeISEhgUqVKjFq1CgKFSpE/fr1+eSTT7j++uspUaIEffv2ZcaMGRQsWJDvvvuOkiVLsmvXLnr27MmhQ4coVaoUo0ePpowv9xaaXEfV6WNYuhS++QbGjYM/PcoA1a8P06c7Vw9hxdvVgmeCCINE4M2UKVPo168f+/bt48knn+Tmm28OajwRnSgefRTcL/pZVr06fPjhxcewadMmRo8ezdChQ9m/fz9vvfUWs2fPpmDBgrz++uv861//4tlnnz3vNUePHiU2Npa33nqLxx9/nFGjRvHMM8/Qv39/4uLiuO+++xg+fDiPPvooEydOvPggTUQZPx46pTOKqWZNqFrVaXYKuU5nX0Yug/erhQhIEABxcXGMHDmSatWq8d133xETk607WnNURCeKUHD11Vef+zbw888/s2nTJurVqwc4t7nVT6eeQIECBWjhFpqpWbMmCxYsADh3VQLQtWtXXnjhhUD8CiYMbNoEr73mdE7Pnu1su+wy58tSo0bOLarFigU1RN+vBryJkGSQlmcRv5iYGMqXL8/TTz/tU/9mIER0osiJK4KLVahQoXOPVZXmzZszduxYr6/x/OOIiooiKaVKmTFpbNzojFr2rGtUu7ZTBK9mzeDFdY6vt55GaALwRXx8PH379qVjx4506dKFvn37BjukC9ic2QFUr1495s2bx073pvSTJ0/yyy+/+Pz6OnXqMGHCBAC+/PJLGjZs6Jc4TWg7cwZef90pcHf99c765Zc7n8kJCc6AuJBJEg88cH6CGDYM5s5N/yeXJYmzZ8/y6aefUrVqVebOnUtCQkKwQ8pQRF9RhJpSpUoxcuRIOnTowBn3K+Abb7xBJW/VzTx88skn9OrVizfffPNcZ7bJXTZtcvoZPL38Mvzzn8GJx6uUK4lhw3JdEsjML7/8QlxcHPPnz6dp06YMHz48pMvuREStp82bN1M55GsDBI6dj/B39KhTYVUVfvgBJkxwrhRSFCvmjJa+7LLgxXiBtH0Qa9Y4d4XMnRu0kELVd999R/fu3Rk8eDDdu3cPSEWFXF/ryZhwd+wY/Pe/sGIFbNniFNpLz5VXwscfwz33+Fae+6L5ejcSXNgHUb260+9gAFi7di1r1qyhW7dutGnThp07d3JZSGX6jFmiMCbI+vZ1Wmc8FSvmfN527OgkhCZNwI81386XldpHnnJxh7Q3CQkJvPbaa7z11lv87W9/o0OHDkRHR4dNkgBLFMYE1L59qbWUliyBxx5LfW7gQLjvPqhQIUh1lNIb2Wwf/hdl8eLF9OrVi82bN9O1a1cGDx4ckCJ+Oc0ShTF+smmT05yUMhfz5MmpYxw8FS4MP/8M1aoFKLCMmpMiaGRzKNi7dy+xsbFceeWVTJ8+/dzYqHBkicIYPxg2zGlSSs9776XeuXTNNfCPf/gxkKxMy2kJIkek3ExSunRpJkyYQJMmTSgSdrVSzmeJwpgc9vzzzjgHgA8+cJqTUhQtGqDyGd4K5FlC8IvDhw/zxBNPMHr0aObPn0+DBg246667gh1WjrBEkQNyosw4wKhRo2jZsiVXXnml32I1OW/XLjhwAN56yym0lzJu6rXXnBIaAeFtjgVLCn43efJk+vfvz4EDBxg0aFDQi/jlNEsUOcCXMuO+GDVqFDVq1LBEESaefRbefPPC7VWqwI8/wt/+FqBAUkZAQ9jMsRBJevbsyejRo6levTrff/89NWrUCHZIOc4ShZ+NGTOGIUOGcObMGerVq8cnn3zC2bNn6dGjB2vWrEFV6dOnD6VKlWLNmjV06NCBAgUKsGzZMtauXcvAgQM5ceIEJUuW5PPPP6dUqVLB/pVyNVVYu9ZpWkop3Nu3L9x8s5MYGjRwOqf9Lr1bWG0EdMB4FvGrU6cOlSpVYuDAgeTLly/IkflH5CWKi6ktnpFs1hzfsGEDkydP5ueffyZv3rz06dOH8ePHc/XVV3Pw4EHWr18POBOUFCtWjI8//phPPvmE6tWrk5CQwCOPPMKUKVMoUaIE48aN44UXXmD48OE5+7sZn6hCYiKUKOHMIZ1i8ODzb3ENiLRXEHb1EFC//vorDzzwAJ07d6Zr1670yQXnPfISRQj58ccfWb58+bl68qdOnaJs2bLcfvvtbN26lUceeYSWLVvSrFmzC167efNmNm7cSNOmTQFITk62SYqC4K+/4K67nNtcPU2aBHffHaBZ4TLqf7AriIBKKeL3zDPPoKrce++9wQ4pYCIvUYRCbXGXqtKzZ09effXVC55bt24dM2bM4KOPPmLSpEkXXCmoKjfccMO5uShMYBw44JTTWLbMWXre4tq8uVOV9ZlnAtC85G10tF1BBNzWrVuJi4tj4cKFNGvWjGHDhnHVVVcFO6yAibxEEUKaNm1Ku3bteOSRRyhRogSHDh3i5MmTFChQgOjoaO69914qVKhwrv58kSJFOO62a1SpUoW9e/eybNkyatWqxZkzZ/jll1+omrZ0qMkxM2c6ySA9yckBrq1ko6NDytatW9m4cSOff/45Xbt2DUgRv1BiicKPqlWrxosvvkjTpk05e/Ys+fLlY+jQoURFRdGrVy9UFRHh7bffBqBHjx7ExcWd68yeOHEiAwYM4Pjx4yQlJfHEE09YovCT+PjUJDFoEFx7rXP1ULy4U4jPr58LGV09WHIIqtWrV7NmzRp69OhB69at2blzJ8WCPk1gkKhqWP3UrFlT09q0adMF23IzOx++SU5Wff551SJFVJ3uatWbbgpwEMOGpb55bKzzM2xYgIMwnk6dOqWDBg3SqKgoLVeunJ46dSrYIeUIYIVm83PXrihMrnLgAJw4AVu3gmfpnaJFncl/Hn/cT2+cWX0l65gOCYsWLaJXr15s3bqVHj168P7774dlEb+cZonCRLyvv3ZGTW/c6NzimtbBg34s4e2tlEbKujUxhYS9e/fSuHFjSpcuzcyZM9O9GzG3iphEoW57f26nYTZjoT8kJzuzwj3/PGzenFpSA6Bly9S5HcqXh0aN/BzMV18543osIYSsTZs2UaVKFUqXLs2kSZNo3LgxhQMyajJ8RESiiI6O5tChQxQvXjxXJwtV5dChQ7n6UvnQIWdQXIqqVeGSS+Cjj+CWWwIYSMqVhE0HGrL+/PNPHn/8ccaMGcO8efNo2LAhd955Z7DDCkkRkSjKlCnDnj17OHDgQLBDCbro6OhcNTDv+HFnQNzmzc4803/9lfrc6tXOZ3RApdfUZNOBhpxJkybx4IMPcujQIZ577jlq1aoV7JBCWkQkinz58lGhQoVgh2EC5NQpp69h82aoUyd1e9Gi8NBDzgxxfuuUTsuqtoad7t27M2bMGGrUqMEPP/xA9YB/mwg/EZEoTO5w9iw0bQpz5py//aqrnFLfAWdVW8NGSt+diFCvXj0qV67ME088Qd689hHoC5/OkoiUAGLc1RWqetDH1zUH/gVEASNU9a00z5cDxgDF3H2eUdXpPsZucpGBA+H991PXX30VChaEihWhdesABWE1l8LSrl276NOnD/fffz/dunXLFUX8clqmiUJE2gIfAAsAAYaKyGOqOjmT10UBQ4DbgD3AchGZoqqbPHZ7Hpigqp+KSBVgOnBVtn4TEzFOn4b9+53Hy5bBSy85t7aCM0p65Ur4+98DEIi3ZqWUpV09hKzk5GSGDBnCoEGDyJMnD/d5TjVossSXK4p/Ajer6h8AIlIKmAV4TRRALWC7qu50XzceaAN4JgoFLnUfFwV+8z10E4nGj4dOndJ/7n//c25p9buMxj5YYggbmzdvplevXixevJgWLVowdOhQypUrF+ywwpYviSJPSpJwHQB8KY9WGoj3WN8D1E6zz0vALBF5GCgENE3vQCLSB+gD2D92hEpKcsY2HDvmrDdrBh06OGMibr7ZmTUuYHNNe/Y7WGIIS9u3b2fr1q2MHTuW++67L1ffNp8TfEkU/xWR6UDKNXhHYKYPr0vvXybtaLBOwOeq+r6I1AXGisj1qnr2vBepDgeGA8TExNiIsghUrVpqkpgzJwAD4dJKexVh/Q5hZ+XKlaxdu5aePXty5513smvXLi699NLMX2gy5UuieAK4F6iP8+E/Bpjow+v2AGU91stwYdNSL6A5gKouFpFooASw34fjmwhRowZs2eI8TkyEoNyIYiOow9apU6d4+eWXee+99yhbtiydO3cmOjrakkQO8uW/5EPAf1R1QhaPvRyoJCIVgL04VyJpRx7tBpoAn4tIZSAap2nLRLjkZFiwAN57zxkYB84trgFPEjaCOqzNnz+fuLg4fvnlF3r16sV7772XqysT+Isv/y1LAnNF5HdgPDDJl9tjVTVJRB7CaaaKAkap6kYReQXnFtspOFcrn4nIYzjNUt3VihVFtKFDnVadtNOaL1vmjIcIGBtBHfb27t1LkyZNKFu2LD/++CNNmjQJdkgRS3z9XBaRGkAH4G5gp6pmMBeYf8XExOiKFSuC8dbmIn3zDbRt6zy+5RanJtPjjzsd1Z71mfwqowRhTU1hY/369VSrVg2AadOm0bhxYwoVKhTkqEKfiKxU1ZjM97xQVi7044H/Ab8DduuR8dmqVTBqFAwZ4qw/9xy89loAA7AZ5CLCwYMHeeyxx/jyyy/PFfG74447gh1WruDLgLveOFcSZYBJwMOqus7fgZnwpgo//eQMllu4MHV7v34BThJwfh+EJYiwo6r85z//4aGHHuLw4cO8+OKL1K6d9k5740++XFFci1Naw9p7TIb++stJACtWOBVdlyw5//nRo6F9e6fsRkB4XkVYR3VY69atG2PHjiUmJobZs2efa3YygZNhohCRQqp6EnjFXT/vXjNVPebn2EwYmDYNdu6EN96APzyGZTZp4kw7+umnzmd0wBIEXDhornp166gOM55F/GJjY7nhhht49NFHrYhfkHg76xOBFsBGnDuSPAfQKdZPkaupwqOPOhMCeTpzBvLlC05MwPlJwgbNhaWdO3fSu3dv7r//fnr06EGvXr2CHVKul2EpDlVt4S7Lqmo5d5nyY0kiFztwAPLkSU0SU6c6M8udPh3EJDF8uDOc25JE2EpOTubDDz+kWrVqLF++nDx5fKkUZALBl87sWaraLLNtJneYOxcaN05d378frrgiaOHY7a4RYtOmTfTs2ZOlS5fSqlUrhg4dmqtmagx13voo8uOMlC4lIkVIbXq6FGt2ynVUoX9/Z8AcwB13wJQpENRaa1bAL2Ls2rWLHTt28NVXX9GxY0cr4hdivF1RPAg8jjMyeyOpieIYMNTPcZkQc8898O23zuOWLZ3mpqCxAn4RYfny5axZs4bevXvTqlUrdu7cSZEiRYIdlklHholCVT8APhCRR1X1wwDGZELImTPOKOqUwfB79wZo0qD0WDNTRPjrr7/45z//yQcffED58uXp0qUL0dHRliRCmLemp1hVnQfsFJELJpt0azWZCKUK994Lkyalblu7NshJwpqZwt7cuXOJi4tjx44dPPDAA7z99ttWxC8MeGt6ug2Yh1NiPC0FLFFEoKQkuOkm2LAhdVvXrs7n9CWXBDiY9EpvWDNT2NqzZw+33XYb5cuX56effqKx510RJqR5a3p63l12CVw4JtgWLHCSRGwsVK7sjLYuXjzAQaTXxGRXEWFr7dq13HjjjZQpU4bvvvuORo0aUTCgIzDNRVNVrz8481Fc6j4eCiwDmmT2On/91KxZU41/JCerOo1OqgsWBCmIYcNSg4iNddZNWNq/f7926tRJAZ07d26ww8n1cKZ3yNbnri/j4fuo6ici0gynMGA/nGlJa/ojcZngiYpyln//O9SvH4QAbFR1RFBVxo8fz4ABAzh69Cgvv/wydevWDXZY5iL4kihSJqxoAYxW1ZUiYkMmI8ixY3DrranrO3YEOAC73TWidOnShXHjxlG7dm1GjhxJ1apVgx2SuUi+JIq1IjIduAZ4TkQKk5o8TBibPx8GDYKff07dtnYtBPQmFLubKSKcPXsWEUFEaNy4MTVr1mTAgAFEpVymmrDmS6LogdPMtF1V/xKREoBV6Qpjqs68EMOGpW5r3Rq++y7AgVhTU0TYvn07vXv3pkuXLvTs2dOK+EWgTJuQVDUZKAE8JSJvATer6mq/R2b84uhRp6BfSpKYPdtJHAFPEpB666slibCUlJTEe++9R7Vq1Vi9ejX58+cPdkjGTzJNFCLyOvAUsNP9eVJEAj1HmckB48dDsWKp6ytXnt83ETAplV7XrHGamyxJhJ0NGzZQt25dnnzySW6//XY2bdrE/fffH+ywjJ/40vR0J1BDVZMARGQUsAp43p+BmZy1eDF06uQ8fuAB+Ne/gjCALoXn1KQ2oVBY2r17N7/++ivjx4+nffv2VsQvwvk6XVQR4LDHYxMmEhLgssvg1ClnvVWr1AqwAeM5whpsatIwtXTpUtauXUufPn1o2bIlO3fupHDhwsEOywSAL7e5vgOsEpERIjISWAG87d+wTE44fNi5gyklSUyYEKSqrylXECnsSiKsnDx5kscff5y6devyzjvvkJCQAGBJIhfJ9IpCVb8UkTlAbZxS4/9U1b1+j8xkmyqMGwdd3OIrefJAYqKzDLjhw53xEbGxdgURhn766Sd69+7Nzp076devH2+99RaXBK3N0gTkl6TwAAAgAElEQVSLt+qxJYBngIrAeuBtVT0RqMBM9jVoAIsWOY/z5XOanwLehJx2EJ1dQYSdPXv2cPvtt1OhQgXmzZtHw4YNgx2SCRJv3zG/AJKBz3Buj/04IBGZbDt8GK65JjVJrFvnzCcRlCTxwAOpVxJ2+2tYWb3aufu9TJkyTJ06lbVr11qSyOW8NT2VVtWWAO7I7FWBCclkR3IyXH556vqqVVCtWgADsJLgYe+PP/5gwIABTJgwgblz5xIbG0vz5s2DHZYJAV77KNLMlZ3Hc11Vj/k5NpMF06c7y6JF4cABp8nJ79JLDlYSPOyoKuPGjeORRx7hxIkTvPbaa9SrVy/YYZkQ4i1RFOf8ubLxWFegnB/jMlmwdatTggPgm28CmCQ8azRZcghbnTt3Zvz48dStW5eRI0dSuXLlYIdkQoy3iYvKBDIQkz2ffw49ejiPe/QI4EhrK78R1jyL+DVr1oy6devy4IMPWhE/ky4rFx6mVOGVV1KTROfOMGpUgN7c85ZXSxJhZ9u2bTRu3JhR7h9Mjx49rNKr8crXkdkmhEybBnfembq+aZMzbWlAeDY52S2vYSUpKYnBgwfz4osvEh0dTYECBYIdkgkTdkURhtq3d5Y33QTr1wcpSViTU1hZt24dderU4emnn6ZFixZs2rSJzpbojY98uqIQkeuBlMkxF6jqRv+FZLzp398pyVGhgnMLbEDYDHRhb8+ePcTHx/Of//yHtm3bWhE/kyW+lBl/CJiAc5dTOWCCiPT35eAi0lxEtorIdhF5JoN92ovIJhHZKCJfpbdPbpeQ4Fw5vP02fPqps+2//w3Qm9vgubD1888/M9StAJlSxK9du3aWJEzWqarXH2AdUNhjvTCwzofXRQE7gH8A+YG1QJU0+1QCVgOXueslMztuzZo1NTfZvVvV6bpO/Rk1yo9vOGyYamxs6k/Kmw4b5sc3NTnp+PHjOmDAABURrVixop4+fTrYIZkQAKzQTD5fM/rxpY9CgESP9UTOH1uRkVo406fuVNUzwHigTZp9egNDVPWwm7T2+3DcXKWcO1qlSBGYOBE2b0690ynHeV49pLCriLAya9Ysrr/+ej7++GMefPBBVq1aZUX8zEXzpY9iLLBERCa563cDY3x4XWkg3mN9D04FWk/XAIjIIpwrkJdU9Ye0BxKRPkAfgHLlcs84v6QkZ5k3Lxzz9zh466gOe/Hx8bRq1Yqrr76a+fPnU79+/cxfZIwPfJkz+x2cD+m/gFNAX1V9z4djp3fVoWnW8+I0PzUCOgEjRKTYBS9SHa6qMaoac8UVV/jw1uHvyy9TR1i/9JKf38ySRFhbuXIlAGXLlmX69OmsWbPGkoTJUV4ThYhEichaVV2uqoNV9X1VXe7jsfcAZT3WywC/pbPPd6qaqKq7gK04iSNXGzgwdS6JTp3goYf89EYpc1dbkghL+/bt49577yUmJoZ5bnPhbbfdRnR0dJAjM5HGa6JQ1WRgk4iUzsaxlwOVRKSCiOQHOgJT0uzzLdAYzs1/cQ2wMxvvFTFWrYL333cep9yVWrSon94sZeY564cIK6rKmDFjqFKlClOnTuWNN96wIn7Gr3zpoygBbBaRxcDJlI2qeo+3F6lqkntr7Uyc/odRqrpRRF7B6X2f4j7XTEQ24cx98aSqHsrm7xL2Fi+GlP/v06Y581vnOM+KrzZ3dVjq2LEjEyZM4JZbbmHEiBFcd911wQ7JRDhx7prysoNIk/S2q+psv0SUiZiYGF2xYkUw3tqvVFOnKr3tNvj++xyuApt20FxsrLO0iq9hwbOI35gxYzh+/Dj9+/cnT1DmtzXhSERWqmpMtl6bWaIINZGWKI4dg3vugdkeaTdH/0nSSxCWHMLKli1biIuLo3v37sTFxQU7HBOmLiZReJsze56qxorIYc6/W0kAVdXLM3ip8VF8fOo4CYCYmBxuBUo7Z4QliLCSmJjIu+++y8svv0yhQoUoXLhwsEMyuZS3PorG7rJEIALJjW66yVmWKQO7d+fw3NZ2y2tYW7NmDT169GDNmjW0a9eOjz/+mCuvvDLYYZlcytvERWfdZbKI1AGuUdUvRORynDIeuwMUY0TZvx/GjYOffoJDh5x+iPj4zF/nE5u3OmLs27ePffv2MWnSJO65x+t9I8b4XaZ3PYnI88AtwNXAF0AB4CtSq8kaHx05AqVKnb9t7docOHB6/RDW1BR2Fi5cyLp16+jfvz/Nmzdnx44dFCxYMNhhGePT7bHtgJuAVQCquldELvVrVBHqssucZcGC8OuvULgw5MjYKM/xEJYcws7x48cZNGgQQ4YMoVKlSvTq1YtLLrnEkoQJGb4kigRVVRFRABGxv94sSkx0+iEAihd3mp9y5K7GlCsJGw8RtmbOnEmfPn2Ij4/nkUce4bXXXrMifibk+JIovhGRIUBREekB9AICNTtzRChXzkkOAEuXXmSSSK8fIuVKwoSV+Ph47rjjDipWrMjChQttdLUJWZkmClV9W0RaAGeAG4HXVXWG3yOLAMnJTuXXFElJcFHz16e93dWamsKOqrJ8+XJq1apF2bJlmTFjBvXr17f6TCak+TQVqpsYLDlkwalTTl9EisWLczBJ2J1MYen333/nwQcfZPLkycydO5fY2FiaNm0a7LCMyZS3AXdpB9qdxwbcZWzqVGjdOnX9r7+gQIFsHMhud40Iqsrnn3/O448/zunTp3n77be55ZZbgh2WMT7zdkVRAmcU9ovAAZwJjAS4D7AO7Qw89RS8+27qemLi+c1PmcqoD8KamcJW+/btmThxIg0aNGDEiBFcc801wQ7JmCzxpSjgUlWtnWbbElWt49fIMhDKtZ4+/RT693cez54Nt96axQOk7YMASw5hKjk5GREhT548fPHFF5w8eZIHHnjAiviZoPFLrScPKiIdgAnubbIdsvNGkS45OTVJTJuWjSQBqVcS1rwU1jZv3kyvXr3o0aMHvXv3pmvXrsEOyZiL4svXm85AV+CQiBwEuuA0PxmX591N7dtncR6JlFnmGjVKHTRnSSIsJSYm8tprr1G9enW2bt1KUb/NOGVMYPlye+xOwB9T6ESEo0fBc96YL7/M4gE8B8xVr27jIcLU6tWr6d69O+vWraNDhw589NFHlCxZMthhGZMjfKn1dAnQHagKnLvZW1Xtay/w3nuwbx9ccgkcPJjFyYaGD3c6rGNjbVR1mPvjjz84ePAg3377LW3atAl2OMbkKF+anr4ArgLuAJbiFAc87ceYwkJyMnzxBbz2mrN+6JBTu8lnnh3XdhURlubPn8+QIUMAaN68Odu3b7ckYSKSL4niGlUdBJxQ1ZFAc+B6/4YV2n780emT6NbNWb/7bihUKIsHsY7rsHXs2DH69+9PbGwsH330EQkJCQAUyNZgGWNCny+JItFdHhGRykARoLz/Qgp97do5y3r1YNcu+OabLB7As8nJkkRYmT59OlWrVmXYsGE8/vjjrFq1yor4mYjny+2xI0XkMpyBdzNxBtu96NeoQthzzzkd2BUqwKJF2TiANTmFrfj4eNq0acO1117LxIkTqV27duYvMiYCZDrgLtQEe8BdynSlq1alTmXqM6vXFHZUlaVLl1KnjjO+dM6cOdxyyy3kz58/yJEZkzV+GXAnIgO8vVBVP8rOG4azqVOdZadO2UgSYP0SYea3336jX79+TJky5VwRv8aNG2f+QmMijLempyvcZSWgFuB+THIHMM+fQYWis2dTC/09//xFHMj6JUKeqjJy5EgGDhxIQkIC7733nhXxM7laholCVV8AEJGZQHVVPeauvwB8HZjwQkdKifAiRaBKlWwcwLMD24S0du3a8c033xAbG8uIESOoWLFisEMyJqh86cwuz/njJhKACv4JJzT165f6+M8/s3mQlGYn68AOSZ5F/O666y6aNWtG7969rYifMfiWKL4ClorIJJz5Ke4BslqoIqx9/72z3LMniyXD07Jmp5C0YcMG4uLi6NWrF71796ZLly7BDsmYkJLp1yVVfQXoA5zCubLoq6qv+TuwUHHgAMTHQ/36ULp0sKMxOenMmTO8/PLL1KhRgx07dnDZZZcFOyRjQpLX78ciEgWsUtUbgeWBCSm0xMU5ywYNghuHyVkrV66ke/fubNiwgc6dO/Phhx9yxRVXZP5CY3Ihr1cUqpoMbBKRXPldets2mDLFefzGG9k8SEoZ8TVrcioskwMOHTrEkSNHmDp1KuPGjbMkYYwXvrS4lwA2i8hi4GTKRlW9x29RhYhrr3WWb72VzQOknbHOOrKDas6cOaxfv54BAwbQrFkzfvnlF6KjozN/oTG5nC+JIrsfk2HtwQdTHz/1VDYOYKOwQ8bRo0d56qmnGD58ONdddx0PPPAAl1xyiSUJY3zkS2f2bGArcNZ9vAhY4u/Agm3MGGe5Z09q2Y4ssVHYIWHq1KlUqVKFESNGMHDgQFauXGlF/IzJokwThYj0BKYAI9xN5YDv/BlUsP31F5w8CbVrZ+NOJ88+CbsdNqji4+Np27YtxYsXZ8mSJbz77rsULFgw2GEZE3Z8aXoagFPCYymAqm4TkVJ+jSrI3n/fWWZ5ELX1SQSdqrJ48WLq1atH2bJlmTVrFvXq1bMifsZcBF+GnZ5W1TMpK+4tsz4RkeYislVEtovIM172ayciKiLZqmyYU06dgh494J//dNYfeyyLB/Bsbpo7164mAmzPnj20bt2aW265hXnznHJkjRo1siRhzEXy5YpikYg8BUSLSGPgQWBaZi9yE8oQ4DZgD7BcRKao6qY0+xXBuWpZmtXgc1JyMni2Sjz3HFx5ZTYOZM1NAXf27Fk+++wznnzySZKSkhg8eDD169cPdljGRAxfriieAo4DW4BHgNnAcz68rhawXVV3ulck44H0JhR+FXiHIM/D/cMPqY9Pn06dC9snNlYiqNq2bUvfvn25+eab2bBhA4899hhRUT5f+BpjMuFtPopHgQmq+hvwqfuTFaWBeI/1PcB5U4KJyE1AWVWdJiIDvcTSB6eMCOXKlctiGL5JuQV2zhzI0k0x1i8RFElJSeTJk4c8efLQtm1bWrVqRa9evZBs3aJmjPHG2xXF1TjNRT+JSG93OtSsSO9/7Lnp9EQkD/AB8ERmB1LV4aoao6ox/hhB+/77sGmTkyAaNcrCC9OOlbB+iYBYt24ddevW5bPPPgPg/vvvJy4uzpKEMX6SYaJQ1YeBssDrOM1Im0Vkqoh0FpFCPhx7j/v6FGWA3zzWiwDXA3NF5H9AHWBKoDu0ExNhoHst869/ZfHFNlYioBISEnjxxRepWbMmv/76q5XdMCZQVNWnH5xmqlbAauCkj/vvxJm7Ij+wFqjqZf+5QExmx61Zs6bmpBdeUAXVd97J4guHDXNeGBubo/GY9C1btkyrVKmigHbp0kUPHjwY7JCMCSvACvXx8z7tj0+zK4hIZaCj+3MCeNGHBJQkIg8BM4EoYJSqbhSRV9yAp/iazPzlyBF49VXncUqVWJ94NjlZn0RAHD58mBMnTjB9+nRatGgR7HCMyVW8dWZXwEkMndz9vgbuVNVtvh5cVacD09Ns+2cG+zby9bg55c03nWWHDpClqQisySkgfvrpJ9avX88jjzxCs2bN2LZtm5XfMCYIvHVmzwWKAd1UtYqqvpyVJBEOduxwlqNHZ+FFnnNfW5LwiyNHjtC7d2+aNGnCsGHDSEhIALAkYUyQeOvMLq+qT6vq6kAGFCinTsGkSVC0KBQokIUX2tzXfvXdd99RpUoVRo0axVNPPWVF/IwJARczA3RYe+klZ5ml5m67mvCr3bt3c++991K5cmWmTJlCTExQK7oYY1y+jMyOSD//7CxHjvTxBdaB7ReqyoIFCwBnMOWPP/7I8uXLLUkYE0J8ShQikl9EKvo7mEA5fBgWLoS8ec+v7+SVdWDnuN27d9OqVSsaNmx4rohfw4YNrYifMSHGl/koWgHrgf+669VFZLK/A/OneLewyPPPZ/GF1uSUI86ePcu///1vqlatyvz58/noo4+siJ8xIcyXK4pXcGo0HQFQ1TVAWF9dDB3qLOvU8WFnK/iX4+655x4efPBB6taty4YNG3j44YetiJ8xIcyXzuxEVT2Spo6OZrRzqDt6FD51yxvedpsPL/jqKydJVK9ufRMXwbOIX4cOHWjTpg3du3e3+kzGhAFfEsVmEWkP5HEH4T1CGM+Z3aWLs6xRA/Jkdj3leZfT3Ln+Di1irV27lp49e9K7d2/69u1Lp06dgh2SMSYLfGl6egioCZwFJgMJwKP+DMpf/voLpk51Hi/1ZZokGzNxUU6fPs3zzz9PTEwMe/bs4cpszQRljAm2TK8oVPUk8LT7E9aGDHGWd9zh3PHklY2ZuCjLli2jW7dubNmyhW7dujF48GAuv/zyYIdljMkGb7WeJuOlL0JV7/FLRH70+uvO0qexE3Y1cVGOHTvGqVOn+OGHH7j99tuDHY4x5iJ4+179ibtsA/wdGOeudwJ2+DMof/j9d6cjO39+KFnSxxfZ1USWzJo1i40bN/LYY4/RtGlTtm7dauU3jIkA3mo9zVbV2UB1VW2nqpNVdTLQAbglYBHmkKfdhrMPP/Rh55RmJ+OTw4cP06NHD26//XZGjhxpRfyMiTC+dGaXFJGrPNbLAWE1tVhiIowd6zzu3TuTna1UR5Z88803VKlShbFjxzJo0CBWrFhhCcKYCOPL7bFPAAtEZKu7Xgno67+Qcl737s7Sp05sK9Xhs927d9OxY0euv/56pk+fzk033RTskIwxfiDODHmZ7CRSAKjirm5S1VN+jcqLmJgYXbFihc/7q6aOl9i3D0qV8rJzytWEjZvIkKoyf/58YmNjAVi4cCG1a9cmX758QY7MGOONiKxU1WxV2/SpKKCqnlLVle5P0JJEdqTc4dS9u49JAqzJKQO//vorLVq0oFGjRueK+NWvX9+ShDERLuLLjKcU/nv5ZS87eSYJa3K6wNmzZ/nkk0+oWrUqCxcu5OOPP6ZBgwbBDssYEyARPXHR7Nnwxx9w5ZVQrlwGO1mSyNRdd93F1KlTuf322xk2bBjly5cPdkjGmADyKVGISFHgaiA6ZZuq/uyvoHLKTz85y5Rqsemyzut0JSYmEhUVRZ48eejUqRPt2rWjS5cuVsTPmFzIl/koegI/Az8Bb7vLN/wcV4745htn2bJlBjtYmY50rVq1ilq1ajHUzbCdOnWia9euliSMyaV86aN4DIgB/qeqDXAKBP7u16hywJkzsGULFCoE6fa1Wuf1BU6dOsWgQYOoVasW+/bto2zZssEOyRgTAnxpejqtqqdEBBHJr6obReQ6v0d2kT7+2Flm2IltTU7nWbJkCd26dWPbtm307NmT9957j8suuyzYYRljQoAvieJ3ESkGTAVmisifwB/+DevizZjhLDt2TOdJa3K6wMmTJ0lMTOS///0vTZs2DXY4xpgQ4kuZ8dbuwxdEpAlQFPjer1HlgEOHnCan0qXTedIqwwLwww8/sHHjRp544gmaNGnCli1byJ8/f7DDMsaEGF86sxulPHYLBX4DtPNnUDnhwIEMkoRdTXDo0CG6detGixYtGDNmDGfOnAGwJGGMSZcvndmvi8jHIlJARK5w56m419+BXay9e6Fu3TQbc3kHtqoyceJEqlSpwldffcXzzz/P8uXLLUEYY7zypY+iAfAUsBqIAl5R1bF+jeoiHTzoLC/oi83lHdi7d++mc+fO3HDDDcyaNYsbb7wx2CEZY8KAL1cUlwI3AnuAM0ApCfEb6lMG2NWpk86TuazJSVX5yR15WL58eebOncuSJUssSRhjfOZLolgGzFHVpkAt4B/AAr9GdZFeeMFZnte6lAsnI9q1axfNmjWjSZMm54r41atXj7yZ1lo3xphUvnxi3K6quwBU9STQX0Ru9W9Y2TdhgrMsWBCiojyeyEV3OiUnJ/PJJ5/w7LPPEhUVxaeffmpF/Iwx2ebL7bG70qn1dNqvUV2ElGanH39M58lc0uzUpk0bvv/+e1q2bMnQoUNthLUx5qJkmijcWk9PAKWB9cDNwBKgkV8jy6Zly+CKK9K54ynCeRbx69KlC506daJz585Wn8kYc9H8WutJRJqLyFYR2S4iz6Tz/OMisklE1onIbBG5qPrV8fFw8iSULJnmiQjvn1ixYgUxMTF8+umnAHTo0IH77rvPkoQxJkf4kihOp8xql1LrCci01pOIRAFDgBY406h2EpEqaXZbDcSo6g3AROCdrASf1pdfOsuUoRLnRGj/xKlTp3j66aepXbs2Bw4csHkijDF+kWHTk4jkVdUksl/rqRawXVV3uscbD7QBNqXsoKpzPPZfAtyf9V8h1b//7SzTvdspwvonFi9eTLdu3fjll1+Ii4vj3XffpVixYsEOyxgTgbz1USwDalxErafSQLzH+h6gtpf9ewEz0ntCRPoAfQDKZTBVXWIi7NkDIlC8uMcTEXw1cfbsWX788UeaNGkS7HCMMRHMW6K4oIFbVWdn4djpNZBrujuK3I/TDxKb3vOqOhwYDhATE5PuMV591Vne6nnjboRdTUyfPp2NGzfy5JNPcuutt7J582bypTvZhjHG5BxvieIKEXk8oydVdXAmx94DeN6XWQb4Le1OItIUeA6IVdWETI6ZodWrneXkye6GCKrrdPDgQR599FHGjRvHjTfeyCOPPEL+/PktSRhjAsJbZ3YUUBgoksFPZpYDlUSkgojkBzoCUzx3EJGbgGFAa1Xdn/XwU23bBoULQ5EinJ8kwriuk6oyfvx4KleuzIQJE3jxxRdZtmyZFfEzxgSUtyuK31X1leweWFWTROQhYCZO0hnlzo73CrBCVacA7+Iko/+4t3Lu9ugTyZL9+z1ui42Q4n+7d++mW7du3HjjjYwcOZJq1aoFOyRjTC6UpT6KrFLV6cD0NNv+6fE4R6ZSGzkSjhyBbt08NoZpv4SqMnv2bJo2bUr58uWZN28eN998M1Hn1SMxxpjA8db0FDa30sTFOcsWLYIbx8XasWMHTZo04bbbbjtXxK9OnTqWJIwxQZVholDVPwMZSHYluN3ft90Gt99OWI7CTk5OZvDgwVSrVo2VK1cybNgwK+JnjAkZYV9v+uOPnWXz5u6GMBw3ceeddzJjxgzuuOMOPv30U8qUKRPskIwx5hxRTXdYQsiKiYnRFStWnFuvUgU2b4Y//3RntGvUyHli7txghOezM2fOkDdvXvLkycOECRNITk6mY8eOVp/JGOMXIrJSVWOy81pfaj2FrMREJ0nkyZPOtKchbNmyZdSsWZN/uzVH2rdvT6dOnSxJGGNCUlgnim3bnGXXrsGNw1d//fUXTzzxBHXr1uXw4cNcffXVwQ7JGGMyFdaJYtEiZ9mhg7shhDuyFy5cSLVq1Rg8eDC9e/dm48aNtAj327SMMblCWHdmDx/uLGNSWt1CuCM7ZWKhOXPm0CilH8UYY8JAWHdmpzTpn/sVQqwje+rUqWzevJmnnnoKgKSkJPLmDevcbIwJU7myMzslOdSv724IoWanAwcO0LlzZ1q3bs3//d//cebMGQBLEsaYsBS2iSKlWuy5RBECzU6qyldffUXlypWZOHEir7zyCkuXLrUifsaYsBa2X3F//NFZnjf/RJDrO+3evZsePXpw0003MXLkSKpWrRq0WIwxJqeE7RXF2bPOsn59gtrsdPbsWWbOnAlA+fLlWbBgAYsWLbIkYYyJGGGbKFKani65hKA1O/3yyy/ceuutNG/enPnz5wNQq1YtK+JnjIkoYZsoLr3UWeYZEfjpTpOSknj33Xe54YYbWLNmDSNHjrQifsaYiBXWfRRXXUVQribuuOMOZs6cSZs2bfj3v//N3//+94C9tzHGBFrYXlH8/jt0Oh64q4mEhATOuh0jcXFxfP3110yePNmShDEm4oVtojhzBjppYK4mlixZQo0aNRgyZAgA7dq1o3379lbEzxiTK4RlovjsM2fA3aVF8evVxMmTJ3nssceoV68ex48fp1KlSn55H2OMCWVhmSgGD4beDKf8Lv/dErtgwQKqVavGhx9+SL9+/diwYQPNz82OZIwxuUdYdmZv2QIj838FZ/Bbs1NSUhL58uVj3rx5NGzY0C/vYYwx4SDsEkVSkrMsUhiolrPNTt9++y2bN29m0KBBNG7cmI0bN1p9JmNMrhd2TU/79zvLEiVy7ph//PEH7du35+6772bixIlWxM8YYzyEXaI4dcpZlip18cdSVcaOHUuVKlX47rvveP3111myZIkV8TPGGA9h95U5IcGZHztPDqS43bt3ExcXR0xMDCNHjuS66667+IMaY0yECcsriudLZr8I4NmzZ5kxYwbgFPFbtGgR8+fPtyRhjDEZCLtEAdBwb/YG2m3bto1GjRrRsmVL5rmJJiYmxor4GWOMF2GXKEpwgJgTWSvbkZSUxNtvv80NN9zA+vXrGT16tN3yaowxPgq7PorL+dN5kIWriVatWjFr1izuuecehgwZwpVXXumn6IwxJvKIpkw+HSaulSI6v3JNSm2a63W/06dPky9fPqKiopg0aRIAbdu2DUCExhgTekRkparGZOe1Ydf0BO5gOy8WLVpE9erVzxXxa9u2rSUJY4zJprBMFJdEp7/9xIkTDBgwgAYNGnD69GkqV64c2MCMMSYChV0fBaQ/hmLevHl069aN3bt389BDD/HGG29QuHAmlx7GGGMyFZaJIqNZIAoWLMiCBQu45ZZbAhqPMcZEsrBLFJ5zBX3zzTds2bKFZ599ltjYWNavX29jIowxJof5tY9CRJqLyFYR2S4iz6Tz/CUi8rX7/FIRuSqzYxbWE+w7c4Z27drRtm1bJk+efK6InyUJY4zJeX5LFCISBQwBWgBVgE4iUiXNbr2Aw6paEfgAeDuz4x4CKq9dy7Rp03jzzTf5+eefrYifMcb4kT+vKGoB21V1p6qeAcYDbdLs0wYY4z6eCDSRTCai/hW4vkYN1q5dyw4zbaoAAAiLSURBVDPPPEO+fPlyOm5jjDEe/NlHURqI91jfA9TOaB9VTRKRo0Bx4KDnTiLSB0ip15GwcOHCDVbED4ASpDlXuZidi1R2LlLZuUh1bXZf6M9Ekd6VQdph4L7sg6oOB4YDiMiK7I4ujDR2LlLZuUhl5yKVnYtUIrIiu6/1Z9PTHqCsx3oZ4LeM9hGRvEBRSCnmZIwxJhT4M1EsByqJSAURyQ90BKak2WcK0M193A74ScOt+JQxxkQ4vzU9uX0ODwEzgShglKpuFJFXgBWqOgUYCYwVke04VxIdfTj0cH/FHIbsXKSyc5HKzkUqOxepsn0uwq56rDHGmMAKy6KAxhhjAscShTHGGK9CNlH4o/xHuPLhXDwuIptEZJ2IzBaR8sGIMxAyOxce+7UTERWRiL010pdzISLt3b+NjSLyVaBjDBQf/o+UE5E5IrLa/X/SMhhx+puIjBKR/SKyIYPnRUQ+cs/TOhGp4dOBVTXkfnA6v3cA/wDyA2uBKmn26Q8MdR93BL4OdtxBPBeNgYLu4365+Vy4+xUB5gNLgJhgxx3Ev4tKwGrgMne9ZLDjDuK5GA70cx9XAf4X7Lj9dC4aAjWADRk83xKYgTOGrQ6w1JfjhuoVhV/Kf4SpTM+Fqs5R1b/c1SU4Y1YikS9/FwCvAu8ApwMZXID5ci56A0NU9TCAqu4PcIyB4su5UOBS93FRLhzTFRFUdT7ex6K1Ab5QxxKgmIj8LbPjhmqiSK/8R+mM9lHVJCCl/Eek8eVceOqF840hEmV6LkTkJqCsqk4LZGBB4MvfxTXANSKySESWiEjzgEUXWL6ci5eA+0VkDzAdeDgwoYWcrH6eAKE7H0WOlf+IAD7/niJyPxADxPo1ouDxei5EJA9OFeLugQooiHz5u8iL0/zUCOcqc4GIXK+qR/wcW6D5ci46AZ+r6vsiUhdn/Nb1qnrW/+GFlGx9bobqFYWV/0jly7lARJoCzwGtVTUhQLEFWmbnoghwPTBXRP6H0wY7JUI7tH39P/Kdqiaq6i5gK07iiDS+nItewAQAVV0MROMUDMxtfPo8SStUE4WV/0iV6blwm1uG4SSJSG2HhkzOhaoeVdUSqnqVql6F01/TWlWzXQwthPnyf+RbnBsdEJESOE1ROwMaZWD4ci52A00ARKQyTqI4ENAoQ8MUoKt791Md4Kiq/p7Zi0Ky6Un9V/4j7Ph4Lt4FCgP/cfvzd6tq66AF7Sc+notcwcdzMRNoJiKbgGTgSVU9FLyo/cPHc/EE8JmIPIbT1NI9Er9Yisj/4TQ1lnD7Y14E8gGo6lCc/pmWwHbgL6CHT8eNwHNljDEmB4Vq05MxxpgQYYnCGGOMV5YojDHGeGWJwhhjjFeWKIwxxnhlicIElIgUF5E17s8+EdnrsZ7fz+/dVES+zcL+FUXklEd8a0QkKqeOn8Ex8opIsvteG9wKyQWycZzRInKtiOTxrKYqIlEisuBiYjS5jyUKE1CqekhVq6tqdWAo8EHKulvQLdRs9YivuqomB+A9j7vnp5q73jurB1DVHqq6Fef/+DMe25NVtUHOhGlyC0sUJmSIyFQRWenOnRDnbvuHiPwiIpe734Z/FpFbM9o/nWO2cucpWIhHRVERKSwin4vIMneOgjuzEGcdEVnsvm6RiFxQFkNEbhWRte6VwSoRKeR+ux/sXimsF5F23t7HHRC2AKjoHvMp97UbRORhd1sREZnhvteGlGOKyEIRqQ68BRRx4/jCvWI54u4zSUSaecT8pYi0EZECIjLGjXGViDT09dyYCBXs+un2k3t/cCp6DvRYv9xdFgQ2kTqPQl+c0tGDcMpme93f4/mCOLVtrsYphjYJ+NZ97h2go/v4MmAbEJ3m9RWBU8Aa9+cjd3tRIMp93Bx3/g+gqcfxZwC13ceFcUYMdwB+cB9fiVPFs2Sa98wLHPn/9u4mRKcojuP496/GyzSMjSSlNFNiChuiGVIaiQ0LRSJLK0UaC6XIa4OUYqHJy+Rlw0IzNaMphcnLbJiJkczCQlmJvCWLn8X/PNyZ7tx5WPDE/1NPc7v33HNPT9M595zz9P+n4yqgA59RLMbzLFTjMa0GgfmpzrOZ+2vT33vAwmx9OfVvANrS8cTUngnAHuBcOt8AvALG/+3/l/j8vU/MKEIl2WlmT4D7eLCyOvgRemAaHm6gZazyGfOAF5KGJAm4nLm2CthrZo+B23hHOSunTdmlpx3p3FTghnkWseN4ZzpSL3AqvflPkS9ZNQFX5Ms/b/DOPC9g4eTUrj48Ic8FYBlwXdJnSR/wOE5NQD+w2syOmlmjpPc59Y2mE2g2sypgLR4v7Wuqtx1A0lM8aFz9L9Qb/jEVGesp/H/Mo98uB5ZI+pKWiiamazXADPxNvAb4VFR+hNFi1BiwTtLQbzT3ENAt6YyZ1eOzhOEPlQ6a2U28A+4zsxXkh3jOU9qj+NlYy0/KJWnQPDruGqDVzDokHS7nIZI+m1kv0IzPTM6XHldmO8N/ImYUoVLUAm9Tp98ALMpca8Xfqg/gUXLHKl/yDE/cMzt1tJsy17qB0gyhFIH3V9r6Oh1vyytgZnWS+iUdwdORzsHTs25Mey3TgUag3Mi2d4D1af+gBt9vuWtmM4GPktqBk3gazB/kSb1KofjzXMNDcC8FejLP2pzum4sP0i/LbGf4B8VAESpFJ1CdlpL2AQ8BzGwlsAA4IekiMM7MtoxWPkueHnY7vl9wl+Ehtven+wfM7Cm+X1KuY/jbe29Bmd1pc7kfeAfcwlP2Psf3GnqAXSozLLykR8BVfDnqAb4vMYB/N31pqaoFyJtNtAH9ZnYp51oXHn67S9K3dO40MMnMBvDluq2qzF+khT8koseGEEIoFDOKEEIIhWKgCCGEUCgGihBCCIVioAghhFAoBooQQgiFYqAIIYRQKAaKEEIIhb4D8xLGbonsPugAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "[mean: 0.65948, std: 0.01181, params: {'learning_rate': 1},\n",
       " mean: 0.70330, std: 0.01894, params: {'learning_rate': 0.5},\n",
       " mean: 0.71970, std: 0.01719, params: {'learning_rate': 0.25},\n",
       " mean: 0.72981, std: 0.01250, params: {'learning_rate': 0.1},\n",
       " mean: 0.72641, std: 0.01263, params: {'learning_rate': 0.05},\n",
       " mean: 0.70386, std: 0.01499, params: {'learning_rate': 0.01}]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'learning_rate': 0.1}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "0.7298057627523985"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "t0 = time()\n",
    "param6 = {'learning_rate':[1, 0.5, 0.25, 0.1, 0.05, 0.01]} \n",
    "grid6 = GridSearchCV(estimator = GradientBoostingClassifier(n_estimators = 90,\n",
    "                                                            max_depth=3,\n",
    "                                                            max_features = 24,\n",
    "                                                            subsample = 0.8,\n",
    "                                                            min_samples_leaf= 65,\n",
    "                                                            min_samples_split = 1800,\n",
    "                                                            random_state = 10), \n",
    "param_grid = param6, scoring = 'roc_auc', n_jobs = 4, iid = False, cv = 5) \n",
    "grid6.fit(x_train, y_train) \n",
    "\n",
    "print(\"Tempo de execução:\", round(time()-t0, 3), \"s\")\n",
    "draw_roc_score(grid6, x_train, y_train, x_test, y_test)\n",
    "display(grid6.grid_scores_, grid6.best_params_, grid6.best_score_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "O valor ideal para learning_rate é 0.1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Passo 7\n",
    "\n",
    "Encontrar o valor ideal para **subsample**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tempo de execução: 15.97 s\n",
      "Train Accuracy: 0.9196%\n",
      "Train Score Roc: 80.705%\n",
      "Train CV Score : Mean - 0.7284231 | Std - 0.0139116 | Min - 0.7119604 | Max - 0.7480388\n",
      "------------------------------------------------------------------------------------------------\n",
      "Test Accuracy: 0.918%\n",
      "Test Score Roc: 73.712%\n",
      "Test CV Score : Mean - 0.7195364 | Std - 0.02954495 | Min - 0.6735364 | Max - 0.7577225\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYoAAAEWCAYAAAB42tAoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzs3XmcjXX7wPHPZRChFFHZk8qWyb5liJSlJLKVfRAtKiptT7tSPdVTFBOqnyeVtKCIJ1mTtcaeJbJFWQsxzMz1++O+jznGLGeWs871fr3mde5zzn3uc81tnOt87+/3e31FVTHGGGPSky/YARhjjAltliiMMcZkyBKFMcaYDFmiMMYYkyFLFMYYYzJkicIYY0yGLFEYE8JEZI6I3BnsOEzeZokijxKRpiKyRET+EpFDIvKDiNQLckx9RCRJRI6JyN8islpE2gczpqwQkbFu7MdE5JSInPa6Pys7x1TV1qr6UW7H6isRecrrdzjp9e9zTERW5+C47UXkl9yM1fiPJYo8SEQuAL4G3gYuBsoAzwIJufw+Udl42Y+qWhQoDrwDfCIixXMzLn9R1btVtagb/0jgU899VW2Ten8RyR/4KLNGVZ/3+p3uBRZ5/U61gh2fCQxLFHnTVQCq+rGqJqnqCVWdo6prPDuIyAAR2SgiR0Vkg4jUdh+vKiLzReSIiKwXkVu9XvOBiLwrIjNF5DjQQkTOE5HXRGSniPzhfusunFmAqpoMTAKKAFW83uNW932PuHFU9XqunIh8ISL7ReSgiIxO69gikk9ERojIr+5+U0TkYve5iiKiItJXRHaJyGERuVtE6onIGvd90zxuZkTkSq9j7wTmuI83EZGl7rHjRaSZ12sWi0gfdztWRBaIyBvuvttEpLXXvmVF5Gu3hbhFRPplEEshEXnd/R3/EJF3RKRQNn+va0VknnuuNojILV7P3S4im9y/o10ico+IlAI+A67yap1cKCL5ReQZEdkuIgdEZJL7pcYEm6raTx77AS4ADgIfAm2Ai1I9fwewB6gHCHAlUAEoAGwFHgcKAjcAR4Gr3dd9APwFNMH5ElIIeBOYjtNyKQbMAF5KJ64+wGJ3Owq4BzgFlHIfuwo4DtzoxvKIG09Bd//VwBs4yaUQ0DSd93kAWAqUBc4DxgEfu89VBBQY6x6jNXAS+AoohdP6+hOIyeQcPwP8N9VjV7rHfh84HygMlHP/LW5yz9nNwAGghPuaxUAfdzsWOA30c3/f+4BdXsf/AaeVWAio7R4nzTiB0cCXwEXu38NM4PlMfqdYYH6qx4oD+4BubkwNgUNAJfdv5whQx923JBDtbrcHfkl1rCeB+cCl7rmZBLwX7P8v9qOWKPLqD1DV/WDfDSS6H+al3edmA0PTeM317odCPq/HPgaecbc/AP7P6zlxP9grez3WCNieTkx93FiOuB+IJ4AuXs8/BUzxup8PJ6E1d4+7H8jvw+++EWjpdf8y9/3yk5Ioyng9fxDo6nX/c+CBTN7jGdJPFOW9HnsCeD/VfnOBO93t1IniF6/9LnCPV9L9YD4NFPF6/lVgfBqx5cNJfhVS/dtuyeR3SitR9AdmpXrsI2CY++9/AOgFFE21T1qJYhfQwOt+FeBosP+v2I8S8tdIjX+o6kacD2ZE5Brgvzjf/rvjfMv9NY2XXY7zDTbZ67EdON+yPXZ5bV+C8815lYh4HhOcb57pWaqqTUWkKDAB5wNsitf77/D6HZJFZJf7/qeBHaqamMGxPSoAX4qI9++RBJT2uv+H1/aJNO4X9eF90uN9jioA3UWko9djBYBv03ntPq/tf9zbojjn5oCqHvd6fgdQI41jXIrTklqd6t8lOyoAN4jIEa/H8gN7VVVFpANOC/Q/IvIT8LCq/pT6IG5/Vhlgjoh4VyotICLFVfVI6teYwLFEYVDVX0TkA2CQ+9AuoHIau/4OlBORfF7Jojyw2ftwXtsHcD5Uq6vqnizGdExEhgC/ishEVf3Zff+ann3E+ZQrh9OqSADKi0h+H5LFLqCfqv6Q+gkRqZiVOLND3a/LXrG8r6qDc3jY34GSIlLEK1mUxzk3qf2Bc0nvalX9I43ns2IXMFNVO6b1pHuO24lIQeBhnNZGVc7+O0FVk0RkL9BWVbM9msr4h3Vm50Eico2IDBORsu79cjgtiaXuLuOB4SJSRxxXikgFYBnOpaRHRKSAiDQHbgE+Set93GTyHvCG24GJiJQRkZt8iVNVD7qx/Mt9aArOh05LESmAc3kjAVgCLAf2Ai+LSBG3s7ZJOoceC7zo/k6IyCXuN99gmAR0FJEbRSTKjbuFiFyelYOo6nZgJTBSnAEE0UBfnA/m1Psm4ZzXN93fXdyO8Nap9/XB50B9EensdkYXFJFG7t9MMRHpIiLFcFp8x3BabuAkq9IiUsTrWGOBUSJSBkBESksYDY+OZJYo8qajQANgmTijk5YC63A+eFHVz4AXgcnuvl8BF6vqKeBWnA7wAzjDV3upakbj4R/F6XBeKiJ/A98BV2ch1jeBtiJyrapuAu7C6bA9gJOkblHVU+6H3y04/QA7cfpeuqZzzP/g9MnMEZGj7u/fIAsx5RpV/Q3oiNP/sh8n9mFk7/9mV5zr+vuAqcDjqjovnX2H4VyaWo4zAGEOXqPLfKWqh3A64ge47/s78BzO5TOAgTitjiM4Hd593cdX4Vxe2+mO4LoQeAlYCCxw/1YWA9dlNSaT++TsVrAxxhhzNmtRGGOMyZDfEoWITBSRP0VkXTrPi4i8JSJbxZnIVNtfsRhjjMk+f7YoPsCZPJSeNjjXRKvgXMd814+xGGOMySa/JQpVXYgzQzM9HXAmZ6mqLgWKi8hl/orHGGNM9gRzHkUZzp54tNt9bG/qHUVkIE6rgyJFitS55pprAhKgMSZvSU6GrIzvUYU//oDERDh+HJKSQLI7ddFLQg7Lc5ZkPxe739MPcYwjQKIzIfOS7BwvmIkirdOZ5j+RqsYBcQB169bVlStX+jMuY0we8OefsG4dzJgBCxfCT+fMF8+aiy+GQ4egR4+cHafF1jga/zaZfAIX+lg3uVAhON+r1GbBpasAONUwhrh9v7PnyqsZ9d3XO9J5eaaCmSh248yq9SiLMwbbGGPO+Osv+O03Z/vnn+HAgXP3+eYbOO88OHYMfnDn20dlUuQ+Kencx+rVg65ds9YqSEyEwYOhWDHfXwNAXBxMnnzu48sXOLcxMVk8oGNPQgKDS5Sga4cO3DlhAve5j4/KQVMnmIliOnCviHyCM9npL1U957KTMSZvmDED9u+Hf/5xtkuUgIMHYc4c349Rvz7UqAFVq8JVV2W8ryqULAmNGjkJIrPEkqvi4mCQWzEndUKIiXGaJQMHZumQqsr48eMZPnw4p0+fpl39+rkUrB8ThYh8jFPVs6SI7Aaexp2tqapjccoat8WZtfsPKTM2jTF5SFISPP88PPvs2Y9HRUHFilCqFLRo4XzTV4WaNeGyNIa9FCmSO/0D2ZZeCyEtC9xWw7hxWU4Iafn1118ZMGAA8+bNo0WLFrz33ntUrpxWubbs8VuiUNXumTyvOOsN5Njp06fZvXs3J0+ezI3DhbVChQpRtmxZChQokPnOxgTB6dOwfLlzyaZ/f/jVq07xF19AnTpQuDBckq1u1wDzTg4LsnDJKJuthvSsXbuWVatWERcXR2xsLJLLGTPsSnik1Zm9fft2ihUrRokSJXL9BIUTVeXgwYMcPXqUSpUqBTsck0ds3w6ffgoFC6b9/Pz5TnJISIB56VSeatMG3n4bcvFLcO5Kr7WQOjnk4od/ZtatW8dPP/1Er169ADh48CAlSpRId38RWaWqdbPzXhFRZvzkyZNUrFgxTycJABGhRIkS7N+/P9ihmAiXnAxLljidyC+/7Ntr6tZ1+g8qV3b6D9q0gfz5oWFDCLkGcOrEkF5rIZdbBr44deoUI0eOZOTIkZQuXZouXbpQqFChDJNETkVEogDyfJLwsPNg/GXTJqeTeezYsy8XATz0EPzrX+n3ERQtCvnCobKcJ0GkTgxBSAhpWbZsGf3792f9+vXcddddvPHGGxQqlK2lzrMkYhKFMcZ327bBHndJo+++c4ad7tnjXD5K68P+xx/PfaxTJ6ePoUEDZw5BRJg8GeLjQyYxeNuzZw/XX389pUuX5uuvv6Zdu3YBe29LFLng4MGDtGzZEoB9+/YRFRXFJW5P3PLlyymY3sVbL3379mXEiBFcfXVWlmowJm2ffZaSCMBJBPv2OaOG/ve/9F/XooVzOSi1G2+Ev/+Ge+5xti+9NPdjDojMRibFx0N0tNOxEiI2b97MVVddRZkyZfj0009p2bIlF1xwQUBjsESRC0qUKEF8fDwAzzzzDEWLFmX48OFn7eNZpDxfOu3v999/3+9xmsj1v//B7+501Tlz0v8srF0bqleHyy+Hbt2gQgVnGGrDhs7s3ojha+dzatHROZ9anUuOHDnCI488wvjx45k/fz7NmjWjY8c0V5z1O0sUfrR161Zuu+02mjZtyrJly/j6669Zs2YNzz33HAkJCVSpUoWJEydSpEgRmjZtyujRo6lRowYlS5bk7rvvZtasWZx//vlMmzaNUqVKsX37dvr168fBgwcpXbo077//PmXLlg32r2mCwDOKaPBgZyTRnjRWxl6yxJl45lGkSAh2GqeWlbkIGQmhzufsmD59OoMHD2bfvn08/PDD1KtXL6jxRHSieOABpyWZHdHR8OabOY9hw4YNvP/++4wdO5Y///yTl19+mblz53L++efz4osv8p///IfHH3/8rNf89ddfxMTE8PLLL/PQQw8xceJERowYwZAhQ4iNjeXOO+8kLi6OBx54gKlTp+Y8SBM2VKFtW/j227Mfv+MOePDBlEtCJUpAgK9O5Ex6ncjZFSYJIS2xsbFMmDCBmjVrMm3aNOrWzdaI1lwV0YkiFFSuXPnMt4ElS5awYcMGGjduDDjD3Jo2bXrOawoXLkybNm0AqFOnDosWLQI40yoB6NWrF0899VQgfgUTJKrO6KLZs515Cu6fwRmDBjmXkO6/H4r7WDwu5KSVIML0Az4nPPPZRIS6detSoUIFHn30UZ/6NwMhohNFbrQIcqpIkSJntlWVm2++mUmTJmX4Gu8/jqioKBITE/0WnwmuQ4fgxAl47z2YMsXpZzjvPKffYG+qymfly8OFF8INN8DIkXD++cGJOcfSm82cBxMEwK5du7j77rvp1q0bPXv25O677w52SOeI6EQRaho3bszQoUPZtm0bV1xxBcePH+f333+nSpUqPr2+YcOGTJkyhe7du/Pf//6XZs2a+Tlik9smT4ahQ536RRs2nPu8CPTs6SQLgCNHnGGoNWo4ndAhKzt1jmJi8nSCSE5OZty4cTz66KMkJSUFraPaF5YoAqh06dJMmDCBrl27curUKQBGjhzpc6IYPXo0/fv356WXXjrTmW1C16FD8O9/Ox/6ixc7l4485ciuvtpJAPv3Q5cuToLo2tXpWwgr2elbyMPJwWPLli3ExsaycOFCWrVqRVxcXGiX3fEM2wyXnzp16mhqGzZsOOexvMzOR/D17q3q9DKc/VO1qupXXwU7uhwYN041Jiblx/OLxcQ4zxmffPXVV1q8eHGdOHGiJicnB+Q9gZWazc9da1EYk4ueegpeeillUZyhQ+H1151tkSCXwc6qtC4nhWhpi3CwevVq4uPj6d27Nx06dGDbtm1cdNFFwQ7LJ5YojMmhrVud8tiPPpryWJs28MILzgS3sOFLITxLDFmWkJDACy+8wMsvv8xll11G165dKVSoUNgkCbBEYUyODBrkfL56++03Z8Zz2AjxQnjh7Mcff6R///5s3LiRXr168frrrwekiF9us0RhTDYcOnR2x/O770LHjlC6dPBiyjKbw+BXe/bsISYmhksvvZSZM2eemRsVjixRGJMFqk5RvLlzUx5bu9YZvhryMrq0ZAki12zcuJGqVatSpkwZpkyZQsuWLSlWrFiww8qRcKgQb0zIePjhlCTx4INO4gibJDFoUEpyACdBjBvnVEq1JJFjhw8fpl+/flSrVu1MNYXbbrst7JMEWIsiV+RGmXGAiRMn0rZtWy4N2xrOke2ZZ5x5EQA7djgzpcOCJ0mAkxgsKeS6L7/8kiFDhrB//34ee+yxoBfxy22WKHKBL2XGfTFx4kRq165tiSIEvfMOPPuss/3ppyGcJDIa0mpJwi/69evH+++/T3R0NN988w21w2qom28sUfjZhx9+yJgxYzh16hSNGzdm9OjRJCcn07dvX+Lj41FVBg4cSOnSpYmPj6dr164ULlyY5cuXs3r1aoYPH86xY8coVaoUH3zwAaXDqrc0PO3f76z6tnQpHDsGEyemPDdwoDOTOmTYkNagUK8ifg0bNqRKlSoMHz6cAiFfxz17Ii9R5KS2eHqyWXN83bp1fPnllyxZsoT8+fMzcOBAPvnkEypXrsyBAwdYu3Yt4CxQUrx4cd5++21Gjx5NdHQ0CQkJDB06lOnTp1OyZEk++ugjnnrqKeJSj8U0uUYV3nrL+RPydumlzupwmzeDj9VW/MMmwIWEHTt2MGjQIHr06EGvXr0YmAfOdeQlihDy3XffsWLFijP15E+cOEG5cuW46aab2LRpE0OHDqVt27a0bt36nNdu3LiR9evX06pVKwCSkpJskSI/+fRTWLUKXn015bEuXZxLTeXKOQv+BJ13P4O1FoIiOTmZd999lxEjRqCq3HHHHcEOKWAiL1GEQm1xl6rSr18/nn/++XOeW7NmDbNmzeKtt97i888/P6eloKpce+21Z0ZPmNy1eTOMGeO0ILxdcQV8+SVce21w4jpH6rkO1s8QFJs2bSI2NpbFixfTunVrxo0bR8WKFYMdVsBEXqIIIa1ataJz584MHTqUkiVLcvDgQY4fP07hwoUpVKgQd9xxB5UqVTpTf75YsWIcPXoUgGrVqrFnzx6WL19O/fr1OXXqFFu2bKF6SNeaDg9PPgkvvphy/7zzYPlyqFYN8gfzf0Rml5as5RA0mzZtYv369XzwwQf06tULCauiXTlnicKPatasydNPP02rVq1ITk6mQIECjB07lqioKPr374+qIiKMGjUKgL59+xIbG3umM3vq1Kncf//9HD16lMTERIYNG2aJIofGjUtJEu+9B927B+nSki/9DZ5tSxBB8fPPPxMfH0/fvn259dZb2bZtG8XDdinBnBFP7324qFu3rq5cufKsxzwzIY3Dzse5fvkF2rd3lhYFePppZ15EQKW3sps3SwpBd/LkSZ577jleeeUVypQpw6ZNm8KyPlNqIrJKVbO1ALe1KEzE27AhZXW4ggXhhx8goOvVp1VTyVoKIemHH36gf//+bNq0ib59+/Lvf/87IpJETlmiMBHr77+d4ax//uncv+suyGS58tyXerSSJYeQtWfPHlq0aEGZMmWYPXt2mqMR86qIqfUUbpfQ/CWvn4djx2DePKhfHy68MCVJvPRSgJNEXBw0b3526QyrqRSSNriLl5cpU4bPP/+ctWvXWpJIJSISRaFChTh48GCe/5BUVQ4ePJgnm8r//AOtWkGxYnDDDbBihfN4x46QmAgjRgQwGO8CfJ7Ce5YgQs6hQ4fo06cP1atXZ+HChQDccsstFC1aNMiRhZ6IuPRUtmxZdu/ezf79+4MdStAVKlQoz03Mi4+H665Lud++PTz2GNSrBwGpqJBeGQ1LECHr888/55577uHgwYM88cQT1K9fP9ghhbSISBQFChSgUqVKwQ7DBNjJk86kOU/9xYoVnVFN+QLVTraV4cJSnz59+PDDD6lduzbffvst0dHRwQ4p5EVEojB5z+efQ+fOKfdHjnRaEQFhK8OFHe8ifo0bN6Zq1aoMGzaM/EGdYRk+fDpLIlIS8AwoXKmqB3x83c3Af4AoYLyqvpzq+fLAh0Bxd58RqjrTx9hNHrF/Pyxb5iw3Om8enDiR8tzQoXD77dCsWQADmjzZud5lCSIsbN++nYEDB3LXXXfRu3fvPFHEL7dlmihEpBPwBrAIEGCsiDyoql9m8rooYAxwI7AbWCEi01V1g9duTwJTVPVdEakGzAQqZus3MRFnwACYMsUZ5uqtYkXn87lePbjttgAHFReX0kk9f36A39xkRVJSEmPGjOGxxx4jX7583HnnncEOKWz50qL4F1BPVf8AEJHSwBwgw0QB1Ae2quo293WfAB0A70ShwAXu9oXA776HbiJVUpJTe2n8eOf+wIFOcrjhBmeiXFRUgANKa0Z1jx4BDsJkxcaNG+nfvz8//vgjbdq0YezYsZQP2dWmQp8viSKfJ0m49uPbsNoywC6v+7uBBqn2eQaYIyL3AUWAVmkdSEQGAgMB+8eOYL/9Bjfd5FR29Rg7NmUqQsDZjOqwtXXrVjZt2sSkSZO4884781wRv9zmS6L4n4jMBDzj/7oBs314XVr/MqknOnQHPlDVf4tII2CSiNRQ1eSzXqQaB8SBU+vJh/c2YUYVvAeudezodFYH7Yu7zagOO6tWrWL16tX069ePW265he3bt3PBBRdk/kKTKV8SxTDgDqApzof/h8BUH163Gyjndb8s515a6g/cDKCqP4pIIaAk8KcPxzdh7ORJKFsWkpOduQ5/ev2LB33epHeSsLkQIe/EiRM8++yzvPbaa5QrV44ePXpQqFAhSxK5yJdEcS/wmapOyeKxVwBVRKQSsAenJZL6++FOoCXwgYhUBQrhXNoyEUgV9u6FJUvAe3EwdzkOjh93Sn8HlSWJsLJw4UJiY2PZsmUL/fv357XXXsuTlQn8zZdEUQqYLyJ7gU+Az30ZHquqiSJyL85lqihgoqquF5HncIbYTsdprbwnIg/iXJbqo3m9DkeEeustZyhraomJQeicTo8libCyZ88eWrZsSbly5fjuu+9o2bJlsEOKWD6vRyEitYGuQEdgm6re7M/A0pPWehQmtG3bBpUrO9u1ajnDXlu1gquugpDpY7QkETbWrl1LzZo1Afj6669p0aIFRUJiYfPQlpP1KLJS7GAX8BuwF7ChRyZTqhAbm5IkRo505qndcw9cfXUIJAlPhdfUVV4tSYSkAwcO0LNnT6699tozRfzat29vSSIAfJlwNwCnJVEW+By4T1XX+DswE9727IEKFZw5EQANGgS4gmt60ltlzkY2hSxV5bPPPuPee+/l8OHDPP300zRokHqkvfEnX/oorsYprWHXe4xPvvgCOnVyti++GNavh0svDW5MwLlDXi05hIXevXszadIk6taty9y5c89cdjKBk26iEJEiqnoceM69f9ZYM1X9O80Xmjxrzx54/XXnB6BmTVi9OkQuMXlPnLPLSyHPu4hfTEwM1157LQ888IAV8QuSjM76VKANsB5nRJL3f3fF+ikM8NdfTpG+xx6DDz5Iefyrr6BDhwAFkXo9iNSsymtY2bZtGwMGDOCuu+6ib9++9O/fP9gh5Xnpdmarahv3tpyqlndvPT+WJPKo5GQnMcyd67QUiheHyy5LSRL33OMsRxqwJAEp1VzT41llzpYiDWlJSUm8+eab1KxZkxUrVpAvYAuLmMz40pk9R1VbZ/aYiVzr1zufxSLw4ovnPj9yJFxwAXTrBiVKBD4+AKKjrZprGNuwYQP9+vVj2bJltGvXjrFjx+a5lRpDWUZ9FAVxZkqXFpFipFx6ugC77JQn7NgBgwfDrFlnP37FFXDXXXDrrc4SpEH74ue55BQf7yQKE7a2b9/Or7/+yuTJk+nWrZsV8QsxGbUo7gEewpmZvZ6URPE3MNbPcZkgW7IEmjRJuT9qVMqSoyFxRSCton0mrKxYsYL4+HgGDBhAu3bt2LZtG8WKFQt2WCYtqprhD/BAZvsE8qdOnTpq/Ovdd1Wd6XKqQ4aonjgR7Ii8jBunGhOTEuC4ccGOyGTR8ePHddiwYZovXz6tVKmSngipP7DIhVM6KVufuxldeopR1QXANhG5NY0EM91/6csE2v79Tqvh3/9OeezNN9OuzxRUtgxpWJs/fz6xsbH8+uuvDBo0iFGjRlkRvzCQ0aWnG4EFOCXGU1PAEkUE+Ocfp/7S1q1nP/7NN9C2bXBiSpctQxrWdu/ezY033kiFChX4/vvvadGiRbBDMj5KN1Go6pPubc/AhWMCadq0lDWn27eHZs3goYdCqJor2DKkEWD16tXUqlWLsmXLMm3aNJo3b875558f7LBMFmRaPdYtFf5/qvq3iIwFagOPqercQASYmlWPzR3r10ONGs52kSKwfTtccklwY0pz4pz3ZDmwy01hZP/+/QwdOpSPP/6Y+fPnE+P5NzRB4e/qsQPdJNEapzDgYOCV7LyZCb5//nGu2tSp49x/7jlnglxIJIlBg1ISg4f3ZDmbMBcWVJWPP/6YatWqMXXqVJ599lkaNWoU7LBMDvhSOMXT5GgDvK+qq0QkFAZImixKSnJaD94efjg4sZxhdZgiTs+ePfnoo49o0KABEyZMoHr16sEOyeSQL4litYjMBK4CnhCRoqQkDxNGvv8+ZXvhQmjaNMgF+9KaC2FJIiwlJycjIogILVq0oE6dOtx///1EhVSHl8kuX/ooooA6wFZVPSQiJYFyqvpzIAJMzfoosichATyjEFetgtq1gxRIWp3T1ooIa1u3bmXAgAH07NmTfv36BTsckw6/9lGoahJQEnhERF4G6gUrSZjs85ThqFMnyEnCux/C0/9gSSIsJSYm8tprr1GzZk1+/vlnChYsGOyQjJ/4UhTwRaAJ4BmO8rCINPEMnzWhb80a6NjR2f744yAFYWtSR5R169bRt29fVq5cSYcOHXjnnXe4/PLLgx2W8RNf+ihuAWqraiKAiEwEfgIsUYSBKVOga1dnu25dqFIlwAFYZ3VE2rlzJzt27OCTTz6hS5cuVsQvwvm6XFQx4LDXtgkDDRvCsmXO9muvwbBhAXzz1AnCOqvD3rJly1i9ejUDBw6kbdu2bNu2jaJFiwY7LBMAvgxzfQX4SUTGi8gEYCUwyr9hmZz417+c0UyeJDFjRoCTBJxdk8kWDQprx48f56GHHqJRo0a88sorJCQkAFiSyEMybVGo6n9FZB7QAKfU+L9UdY/fIzPZ8ttv8Pzzzvbdd8Ojj0LFigEOwmoyRYzvv/+eAQMGsG3bNgYPHszSevTpAAAgAElEQVTLL7/MeeedF+ywTIBlVD22JDACuBJYC4xS1WOBCsxk3axZKYX8Xn7ZSRIB591pbTWZwtru3bu56aabqFSpEgsWLKBZs2bBDskESUaXnv4PSALewxke+3ZAIjLZct11KUli+HB45JEgBGEjmyLCzz87o9/Lli3LjBkzWL16tSWJPC6jRFFGVR9V1W+AIYCtNRmCNm50+iPi453706bBq68GYca1JYmw98cff9C1a1dq167NAncQws0330zhwoWDHJkJtgz7KFKtlZ3P+76q/u3n2EwmVKFaNWf7ssucZFGqVBACsSQR1lSVjz76iKFDh3Ls2DFeeOEFGjduHOywTAjJKFGU4Oy1svG6r0B5P8ZlfPD++85tdDT8HIy58jZHIiL06NGDTz75hEaNGjFhwgSqVq0a7JBMiMlo4aKygQzEZM26ddC/v7P90UdBCMAK+oU17yJ+rVu3plGjRtxzzz1WxM+kydcJdyaEPPKI0w8B0LdvyuUnv7OCfhFh8+bNDBgwgF69etG/f3/69u0b7JBMiLN1JcLMypUpSWLyZJg4MYBv7plEB1bQLwwlJibyyiuvUKtWLdasWWOd1MZn1qIII/Png2c9+uefh+7dA/Cm3q2I+HinQ8Qm0YWdNWvW0K9fP1atWkXHjh0ZM2YMl112WbDDMmHCp0QhIjWApu7dRaq63n8hmbT8+mtKknjpJRgxIgBvmrofIjraJtGFqd27d7Nr1y4+++wzOnXqZEX8TJZkeulJRO4FpuCMcioPTBGRIb4cXERuFpFNIrJVRNL8aBORLiKyQUTWi8jktPbJ66ZOhSuvdLabNAlCkrA1q8PSkiVLGDt2LMCZIn6dO3e2JGGyzJcV7tYAjT3lO9ylUJeo6rWZvC4K2AzcCOwGVgDdVXWD1z5VcJLQDap6WERKqeqfGR03L61wd/KkUxZ8927n/sCBzmd2rvO+vORhndVh69ixYzzxxBO8/fbbVK5cmXXr1ll9JuPfFe5w5k2c9rp/mrPnVqSnPs7yqdtU9RTwCdAh1T4DgDGqehggsySR11xxRUqSmDPHT0kCzu6k9rDO6rA0Z84catSowdtvv80999zDTz/9ZEnC5JgvfRSTgKUi8rl7vyPwoQ+vKwPs8rq/G6cCrberAETkByAKeEZVv019IBEZCAwEKF8+b8zz++kn2LvX2T50CC66KJffwDqpI86uXbto164dlStXZuHChTRt2jTzFxnjA1/WzH4F50P6H+AEcLeqvubDsdNqdaS+zpUfqAI0B7oD40WkeBoxxKlqXVWte8kll/jw1uGvTh3n9ssv/ZQkvNeutk7qsLZq1SoAypUrx8yZM4mPj7ckYXJVZrWeooCfVLUWTh9DVuwGynndLwv8nsY+S1X1NLBdRDbhJI6svldEmT49Zfu22/zwBp6WhF1aCmv79u3jvvvuY+rUqcyfP5+YmBhuvPHGYIdlIlCGLQpVTQI2iEiZbBx7BVBFRCqJSEGgGzA91T5fAS3gzPoXVwHbsvFeEeO116CD25Pz9de5fPC4OGjePGXlOUsSYUlV+fDDD6lWrRozZsxg5MiRVsTP+JUvfRQlgY0i8iNw3POgqt6e0YtUNdEdWjsbp/9hoqquF5HngJWqOt19rrWIbMBZ++JhVT2Yzd8l7I0ZAw8/7Gy/8gq0a5cLB02r7IanNpMJS926dWPKlCk0adKE8ePHc8011wQ7JBPhfBke2zKtx1V1rl8iykSkDo9duhQaNXK2p02DW2/NpQN7WhDR7nIiVrwvLHkX8fvwww85evQoQ4YMIV8+q8JjfJOT4bG+rJkdlISQl7z0Ejz+uLP91lu5mCQ8bERTWPvll1+IjY2lT58+xMbG0rt372CHZPKYdL+OiMgC9/awiBzy+jksIocCF2Lk+v13uPPOlCTx6qtw3325+AZxcSmXm0zYOX36NCNHjqRWrVps2LCBokWLBjskk0dl1KJwKwtRMhCB5DVr18K1XnPbv/kmZc3rbEs9w9qTJKw/IuzEx8fTt29f4uPj6dy5M2+//TaXXnppsMMyeVRGCxclu7dJItIQuEpV/09ELgaKAjsDFGPEOXYsJUm0bg3/939QunQODph6pbmYmJRb65MIS/v27WPfvn18/vnn3H57huNGjPE7XzqznwSaAJVV9Sp3qOynqhqUGT3h3pl99ChccEHK/UxOv2+8O6wtMYStxYsXs2bNGoYMcWpu/vPPP5x//vlBjspECn/XeuoMtMUdGquqe4ALMnyFSdPo0SlJ4rzznKJ/ucbTYW1JIuwcPXqUe++9l+uvv54333yThIQEAEsSJmT4kigS1Gl2KICI2F9vFo0YAU2bwjPPOPefeAKOHHGSRbZ5Js95WhMmLM2ePZsaNWrwzjvvMHToUCviZ0KSLxPuvhCRMcCFItIX6A8EcgHOsLZ3L4wa5Ww3b+5Mohs+PBcO7Kn4Gh1ttZrC1K5du2jfvj1XXnklixcvttnVJmT5Mo9ilIi0AU4BtYAXVXWW3yOLAHv3wuWXO9svvOC0JHKFZ9hrTIzNjwgzqsqKFSuoX78+5cqVY9asWTRt2pRChQoFOzRj0uXTtE5VnaWqD6rqA5YkfLNnT0qSKFgwZa5ErvAMgbVWRFjZu3cvnTp1okGDBixwR6i1atXKkoQJeem2KETkMOeWBT9DVS/2S0QR4I8/oGzZlPsnT0KurT7p3ZqwjuuwoKp88MEHPPTQQ5w8eZJRo0bRpEmTYIdljM8yuvRUEmdNiaeB/TgLGAlwJ2Ad2un47jvwVHquXh3WrMnlJOFZx9paE2GjS5cuTJ06leuvv57x48dz1VVXBTskY7Ik3UtPqpqkqolAa1V9S1UPq+ohVX0bZ5U7kwZPkrj/fli1CnK1ZputIxE2kpKSSE5OBuCWW27hnXfeYf78+ZYkTFjy5WNMRaSriPO9WES6+jmmsHTqVEoJjgoV4D//yeHw19TsklPY2LhxI9dffz0TJkwAoFevXgwePNgqvZqw5ctfbg+gF3BQRA4APXEuPxnX3r1OUpjldvMvXpzLb2CXnMLC6dOneeGFF4iOjmbTpk1ceOGFwQ7JmFzhy/DYbUBuLKETsTyjmwAOH4bi56z6nUXpFfezS04h6+eff6ZPnz6sWbOGrl278tZbb1GqVKlgh2VMrsg0UYjIeUAfoDpwZhyfqtonFjByZMp2rtRt8m49WHG/sPHHH39w4MABvvrqKzp41rI1JkL4MjP7/3DWsW4PvIhzKWq9P4MKFwMGwPjxzva0ablwQO8kYa2HkLdw4ULWrl3LPffcw80338zWrVspXLhwsMMyJtf50kdxlao+BhxT1QnAzUAN/4YV+uLiUpLEjz/mcFU6T90mSxJh4e+//2bIkCHExMTw1ltvnSniZ0nCRCpfEsVp9/aIiFQFigEV/BdS6Nu0KeUz/fPPoWHDHBzM04rwjGiyJBHSZs6cSfXq1Rk3bhwPPfSQFfEzeYIvl54miMhFOBPvZuNMtnvar1GFqB07nHkSW7Y49z/+GLK9pkzqxYYsQYS8Xbt20aFDB66++mqmTp1KgwYNgh2SMQHhy6ince7mPKC8f8MJXfv3Q8WKKfdffhm6dcvBAT3VX62jOqSpKsuWLaNhw4aUK1eOOXPm0KRJEwoWLBjs0IwJmIxqPd2f0QtV9a3cDyd09e3r3DZpkgvzJKz6a1j4/fffGTx4MNOnT2f+/PnExMTQokWLzF9oTITJqEVxiXtbBagPzHDvtwcW+DOoUJOQAN9842wvWpQLB7TqryFNVZkwYQLDhw8nISGB1157zYr4mTwt3UShqk8BiMhsIFpV/3bvPwV8GpjwQoOnCnTLlrlQ4M9KcYS8zp0788UXXxATE8P48eO58sorgx2SMUHlS2d2BcB7decEoJJ/wgk927albM+ZkwsHtNZESEpKSkJEyJcvH7fddhutW7dmwIABVp/JGHwbHjsZWCYiT4rIE8AS4L/+DSt0VKvm3L7xRg4rwXrmSng6sK01ETLWrVtHkyZNzhTx69mzJ4MGDbIkYYzLl1FPz4nILKCZ+9DdqrrCv2EFX3Kys/iQO5eKoUOzeaDUw2A9o5xM0J06dYqXXnqJF198kQsvvJCLLroo2CEZE5IyTBQiEgX8pKq1gIhPDt5693aqwgJs2JCDvgkbBhuSVq1aRZ8+fVi3bh09evTgzTff5JJLLsn8hcbkQRkmClVNEpENIlJGVfcEKqhg++47+K97cS0hwVnzOltsGGzIOnjwIEeOHGHGjBm0b98+2OEYE9J86cwuCWwUkR+B454HVTW7c5JDmircdJOzPX58DpOErSERUubNm8fatWu5//77ad26NVu2bKFQoUKZv9CYPM6XRPGy36MIIW3bOv0TAP37Z/MgVgU2pPz111888sgjxMXFcc011zBo0CDOO+88SxLG+CjTYR2qOhfYBCS72z8AS/0dWLB8+61ze+RIDg5ia1uHjBkzZlCtWjXGjx/P8OHDWbVqlRXxMyaLfFm4qB9wL3AhUBmn3tM7QCv/hhZYiYnQzB3XdccdkO1VLG1CXcjYtWsXnTp14pprruGrr76iXr16wQ7JmLDky0Dx+4GGwN8AqroZKO3PoIJhwQJnXQmA55/PwYFsQl1QqSpLliwBOFPEb+XKlZYkjMkBXxLFSVU95bnjDpn1iYjcLCKbRGSriIzIYL/OIqIiUtfXY+emgwehlds+WrsWrr46mwey1kRQ7d69m1tvvZUmTZqwwJ230rx5c6v0akwO+ZIofhCRR4BCItICp87T15m9yE0oY4A2QDWgu4hUS2O/YjitlmVZCTy3TJsGJUs62+XLQ/Xq2TyQjXIKmuTkZMaNG0e1atWYO3cur7/+Ok2bNg12WMZEDF8SxSPAUeAXYCgwF3jCh9fVB7aq6ja3RfIJkNaq888Dr3B2PamA+dQtbzh8OGzenMOJdWAd2EHQqVMn7r77burVq8e6det48MEHiYryueFrjMlERutRPABMUdXfgXfdn6woA+zyur8bOGtJMBG5Diinql+LyPAMYhkIDAQoXz53106aMsW5ffXVHBzELjkFXGJiIvny5SNfvnx06tSJdu3a0b9/fyTH5X2NMall1KKoDKwQke9FZIC7HGpWpPU/Vs88KZIPeAMYltmBVDVOVeuqat3cLLPw+OOQlAR1s9sz4in0Z5ecAmrNmjU0atSI9957D4C77rqL2NhYSxLG+Em6iUJV7wPKAS/iXEbaKCIzRKSHiBTx4di73dd7lAV+97pfDKgBzBeR33BGVk0PVIf233/DSy852//5TzYO4OmT8LQk7JKT3yUkJPD0009Tp04dduzYYbWZjAmQzGo9JeP0ScwVkfzATcALwHtAZsliBVBFRCoBe4BuwJmv3Kr6F055EABEZD4wXFVXZv3XyLr9+53bZ5+Fxo2zcQDrkwioFStW0KdPHzZs2EDPnj154403KFGiRLDDMiZP8KWEByJSFeeDvhtwDHg6s9eoaqKI3AvMBqKAiaq6XkSeA1aq6vTsh51zGzY4t9XOGYflA+uTCLjDhw9z7NgxZs6cSZs2bYIdjjF5Skad2ZVwEkN3d79PgVvcCXc+UdWZwMxUj/0rnX2b+3rc3PDGG85tlhOFDYMNmO+//561a9cydOhQWrduzebNm638hjFBkFFn9nygONBbVaup6rNZSRKh7tgxOP/8LCYKK/YXEEeOHGHAgAG0bNmScePGkeCuHmVJwpjgyKgzu4KqPqqqPwcyoEBISoIVK+DGG7P4QuuX8Ltp06ZRrVo1Jk6cyCOPPGJF/IwJAT71UUSa8eOd2+LFs/Fi65fwm507d3LHHXdQtWpVpk+fTt1sj1s2xuSmPLd6/IkTcM89zrann8Inng5sk6tUlUWLFgHOZMrvvvuOFStWWJIwJoT4lChEpKCIXOnvYAJh8mTn0lOpUlloUVgHtl/s3LmTdu3a0axZszNF/Jo1a2ZF/IwJMZkmChFpB6wF/ufejxaRL/0dmD+oQmyss/3LL1mo62R9E7kqOTmZd955h+rVq7Nw4ULeeustK+JnTAjzpY/iOZwaTfMAVDU+XFsXnvlZpUrBRb4WJLE5E7nu9ttvZ9q0adx4443ExcVRsWLFYIdkjMmAL4nitKoeSVVHR9PbOVSpwuHDzvaOHT6+yC455RrvIn5du3alQ4cO9OnTx+ozGRMGfOmj2CgiXYB8IlJJRN4kDNfMfv1157ZvXyhUyMcX2SWnXLF69WoaNGhAXFwcAN27d6dv376WJIwJE74kinuBOkAy8CWQADzgz6D8IT7euc3yMqd2ySnbTp48yZNPPkndunXZvXs3l156abBDMsZkQ6aXnlT1OPCo+xOWEhLgv/91WhJlyvjwgrg4pzURHw/R0X6PLxItX76c3r1788svv9C7d29ef/11Lr744mCHZYzJhoxqPX1JBn0Rqnq7XyLyA88EuxYtfHyBd5Kwvols+fvvvzlx4gTffvstN910U7DDMcbkQEYtitHubQfgcuAj93534Fd/BpXbprt1aidNysKLoqNh/nx/hBOx5syZw/r163nwwQdp1aoVmzZtsvIbxkSAjGo9zVXVuUC0qnZW1S9V9UugK9AkYBHm0OjRMGeOMyTWp+ULbAZ2lh0+fJi+ffty0003MWHCBCviZ0yE8aUzu5SIVPS6Xx4Im6XFHnC73d9+28cXeEY62SUnn3zxxRdUq1aNSZMm8dhjj7Fy5UpLEMZEGF/mUQwDFonIJvd+FeBu/4WUe/btc8p11KsHXbr48AKbXJclO3fupFu3btSoUYOZM2dy3XXXBTskY4wf+DLq6RsRuQrwrNywQVVP+Des3DFunHP74IM+7GyT63yiqixcuJCYmBjKly/P999/T4MGDShQoECwQzPG+IlPRQFV9YSqrnJ/wiJJACxc6Nx26+bDzja5LlM7duygTZs2NG/e/EwRv6ZNm1qSMCbCRXSZ8e+/d259ngBsl5zSlJyczOjRo6levTqLFy/m7bff5vrrrw92WMaYAInoRAFwww0+7GQjnTJ02223cd9999G0aVPWr1/PvffeS758Ef+nY4xx+bTCnYhcCFQGzlRJUtUl/goqN/zvf85t1ao+7Gwjnc5x+vRpoqKiyJcvH927d6dz58707NnT6jMZkwf5sh5FP2AJ8D0wyr0d6ee4cuypp5zbXr0y2dFGOp3jp59+on79+owdOxZwivj16tXLkoQxeZQv1w8eBOoCv6nq9TgFAvf6NaocSkqCZcuc7fr1M9nZWhNnnDhxgscee4z69euzb98+ypUrF+yQjDEhwJdLTydV9YSIICIFVXW9iFzj98hy4NVXndtMRztZa+KMpUuX0rt3bzZv3ky/fv147bXXuMjn1Z2MMZHMl0SxV0SKAzOA2SJyCPjDv2HljKemk2cNinRZa+KM48ePc/r0af73v//RqlWrYIdjjAkhour7YnUi0hK4EPhGVRP8FlUG6tatqytXrsxwn2LFnKVOd+7M5GDNmzu3ebT437fffsv69esZNmwYAKdOnaJgwYJBjsoY4w8iskpV62bntb50Zjf3bLuFAr8AOmfnzQIhIQGOHctCSfE86ODBg/Tu3Zs2bdrw4YcfcurUKQBLEsaYNPnSmf2iiLwtIoVF5BJ3nYo7/B1Ydv3hXhSrUCG4cYQiVWXq1KlUq1aNyZMn8+STT7JixQpLEMaYDPmSKK4H9gA/4wyT/UJVb/NrVDnw5ZfO7ZVXZrJjHpxkt3PnTnr06EG5cuVYuXIlzz//vFV6NcZkypdEcQFQC9gNnAJKSwgPqN+927mNiclkxzzSka2qfO/WMqlQoQLz589n6dKl1KpVK8iRGWPChS+JYjkwT1VbAfWBK4BFfo0qBz74wLnNcApAHhkWu337dlq3bk3Lli3PFPFr3Lgx+fP7NCHfGGMA34bH3qSq2wFU9TgwRER8qaAUcMnJcOAAVKkC6ZYiygPlxJOSkhg9ejSPP/44UVFRvPvuu1bEzxiTbb6sR7E9jVpPJ/0aVTbFxzu3XbtmsFMeKCfeoUMHvvnmG9q2bcvYsWNthrUxJkcyTRRuradhQBlgLVAPWAo092tk2fD1185tjRqZ7BiBl5y8i/j17NmT7t2706NHD6vPZIzJMb/WehKRm0Vkk4hsFZERaTz/kIhsEJE1IjJXRHI0qPXnn53bjh1zcpTws3LlSurWrcu7774LQNeuXbnzzjstSRhjcoUvieKkZ1U7T60nINNaTyISBYwB2uAso9pdRKql2u1noK6qXgtMBV7JSvCprVoFhQtDutMCImxI7IkTJ3j00Udp0KAB+/fvp4JNHjHG+EG6l55EJL+qJpL9Wk/1ga2qus093idAB2CDZwdVnee1/1Lgrqz/Co6tW2HXLrj88nR2iLBO7B9//JHevXuzZcsWYmNjefXVVylevHiwwzLGRKCM+iiWA7VV9Vb3/lPetZ58OHYZYJfX/d1Agwz27w/MSusJERkIDAQoX758mi+e56acZ59N5+gR1ol94sQJkpOT+e6772jZsmWwwzHGRLCMEsU5F7hVdW4Wjp3WBfI0KxCKyF04/SBpTpNT1TggDpyigGnt88QTzm379mk8GSHzJmbOnMn69et5+OGHueGGG9i4cSMFChQIdljGmAiXUaK4REQeSu9JVc2siPduwHtcZlng99Q7iUgr4AkgJicVaffvd24vvTSNJ8N8FvaBAwd44IEH+Oijj6hVqxZDhw6lYMGCliSMMQGRUWd2FFAUKJbOT2ZWAFVEpJKIFAS6AdO9dxCR64BxwK2q+mfWw3f8/bdz26dPBjuFYWtCVfnkk0+oWrUqU6ZM4emnn2b58uVWxM8YE1AZtSj2qupz2T2wqiaKyL3AbJykM9FdHe85YKWqTgdexUlGn7lDOXd69Yn47O67nds0B/14X3YKMzt37qR3797UqlWLCRMmULNmzWCHZIzJg7LUR5FVqjoTmJnqsX95befKUmqe+RPu+jtnC7PLTqrK3LlzadWqFRUqVGDBggXUq1ePqKioYIdmjMmjMrr0FBZDaVThl1/gllucle3OEmad2L/++istW7bkxhtvPFPEr2HDhpYkjDFBlW6iUNVDgQwkuzwNhjQLooZJayIpKYnXX3+dmjVrsmrVKsaNG2dF/IwxISPs603/+9/O7Sup53SHUWvilltuYdasWbRv3553332XsmXLBjskY4w5I+wThad/4syKdnFxTkvCU6ojRFsTp06dIn/+/OTLl48+ffrQs2dPunXrZvWZjDEhx5daTyHrT3dA7a3e46QmT3bqjcfEhOws7OXLl1OnTh3eeecdALp06UL37t0tSRhjQlJYJ4pHHnFuz5mNHR0N8+eHXJL4559/GDZsGI0aNeLw4cNUrlw52CEZY0ymwv7SE0D//sGOIHOLFy+md+/ebNu2jUGDBjFq1CguvPDCYIdljDGZCutE8eGHUKZMBsuehhDPwkLz5s2jefPmwQ7HGGN8FgYfsWk7ccK5TU72ejDE1puYMWMGr7jDsVq0aMGGDRssSRhjwk7YJootW5zbwYPdB0JovYn9+/fTo0cPbr31Vj7++GNOnToFQP40J3sYY0xoC9tEMXq0c3um/FEIrDehqkyePJmqVasydepUnnvuOZYtW2ZF/IwxYS1sv+IecueNd+jg9WCQJ9ft3LmTvn37ct111zFhwgSqV68etFiMMSa3hG2LIl8+KF4cgj31IDk5mdmzZwNQoUIFFi1axA8//GBJwhgTMcI2USxZApUqBTeGLVu2cMMNN3DzzTezcOFCAOrXr29F/IwxESVsE8WePRCsz+PExEReffVVrr32WuLj45kwYYIV8TPGRKyw7KNITHRuzyx7GuDFidq3b8/s2bPp0KED77zzDpdffnlA3tcYY4IhLBPFpk3O7ZlugACUE09ISKBAgQLky5eP2NhY+vXrxx133GH1mYwxES8sLz1Nd1fePqsB4ccRT0uXLqV27dqMGTMGgM6dO9OlSxdLEsaYPCEsE8WaNc7tDTfg19nYx48f58EHH6Rx48YcPXqUKlWq+OV9jDEmlIXlpSdwhsWedx5+u+y0aNEievfuzfbt2xkyZAgvvfQSF1xwQa6+hzHGhIOwTBSffgo1auDXVewSExMpUKAACxYsoFmzZrl6bGOMCSdhlygSEkAVypYl11sTX331FRs3buSxxx6jRYsWrF+/3uozGWPyvLDrozhwwLm96y73gVxoTfzxxx906dKFjh07MnXqVCviZ4wxXsIuUXjKivc4lvNObFVl0qRJVKtWjWnTpvHiiy+ydOlSK+JnjDFewu4r8/79cMkl5Mplp507dxIbG0vdunWZMGEC11xzTe4EaYwxESTsWhSqTrIAsnXZKTk5mVmzZgFOEb8ffviBhQsXWpIwxph0hF2iKMl+fi3fHOLjs/zazZs307x5c9q2bcsC97JV3bp1rYifMcZkIOwSxcUcovyheIiO9vmyU2JiIqNGjeLaa69l7dq1vP/++zbk1RhjfBR2fRQAp6pGk3/+fJ/3b9euHXPmzOH2229nzJgxXHqmmqAxxpjMhGWiiPIh6pMnT1KgQAGioqIYOHAgAwcOpFOnTv4PzhhjIkzYXXoCKFAg4+d/+OEHoqOjzxTx69SpkyUJY4zJprBMFOkVbT127Bj3338/119/PSdPnqRq1aqBDcwYYyJQWF56SitPLFiwgN69e7Nz507uvfdeRo4cSdGiRQMemzHGRJqwTBTpOf/881m0aBFNmjQJdijGGBMxwi5RFOPYme0vvviCX375hccff5yYmBjWrl1rcyKMMSaX+bWPQkRuFpFNIrJVREak8fx5IvKp+/wyEanoy3H3tWtH586d6dSpE19++eWZIn6WJIwxJveJqvrnwCJRwGbgRmA3sALorqobvPYZAlyrqneLSDego6p2zei4l0shPVG8MCdOnOCZZ55h2Kw7X7oAAAjFSURBVLBhFMhsGJQxxuRxIrJKVetm57X+bFHUB7aq6jZVPQV8AnRItU8H4EN3eyrQUjJZiHofCdSoUYPVq1czYsQISxLGGONn/uyjKAPs8rq/G2iQ3j6qmigifwElgAPeO4nIQMBT/S9h8eLF66yIHwAlSXWu8jA7FynsXKSwc5Hi6uy+0J+JIq2WQerrXL7sg6rGAXEAIrIyu82nSGPnIoWdixR2LlLYuUghIiuz+1p/XnraDZTzul8W+D29fUQkP3AhcMiPMRljjMkifyaKFUAVEakkIgWBbsD0VPtMB3q7252B79VfvevGGGOyxW+Xntw+h3uB2UAUMFFV14vIc8BKVZ0OTAAmichWnJZENx8OHeevmMOQnYsUdi5S2LlIYeciRbbPhd+GxxpjjIkMYVkU0BhjTOBYojDGGJOhkE0U/ir/EY58OBcPicgGEVkjInNFpEIw4gyEzM6F136dRURFJGKHRvpyLkSki/u3sV5EJgc6xkDx4f9IeRGZJyI/u/9P2gYjTn8TkYki8qeIrEvneRGRt9zztEZEavt0YFUNuR+czu9fgSuAgsBqoFqqfYYAY93tbsCnwY47iOeiBXC+uz04L58Ld79iwEJgKVA32HEH8e+iCvAzcJF7v1Sw4w7iuYgDBrvb1YDfgh23n85FM6A2sC6d59sCs3DmsDUElvly3FBtUfil/EeYyvRcqOo8Vf3HvbsUZ85KJPLl7wLgeeAV4GQggwswX87FAGCMqh4GUNU/AxxjoPhyLhS4wN2+kHPndEUEVV1IxnPROgD/p46lQHERuSyz44Zqokir/EeZ9PZR1UTAU/4j0vhyLrz1x/nGEIkyPRcich1QTlW/DmRgQeDL38VVwFUi8oOILBWRmwMWXWD5ci6eAe4Skd3ATOC+wIQWcrL6eQKE7noUuVb+IwL4/HuKyF1AXSDGrxEFT4bnQkTyAW8AfQIVUBD58neRH+fyU3OcVuYiEamhqkf8HFug+XIuugMfqOq/RaQRzvytGqqa7P/wQkq2PjdDtUVh5T9S+HIuEJFWwBPAraqaEKDYAi2zc1EMqAHMF5HfcK7BTo/QDm1f/49MU9XTqrod2ISTOCKNL+eiPzAFQFV/BArhFAzMa3z6PEktVBOFlf9Ikem5cC+3jMNJEpF6HRoyOReq+peqllTViqpaEae/5lZVzXYxtBDmy/+Rr3AGOiAiJXEuRW0LaJSB4cu52Am0BBCRqjiJYn9AowwN04Fe7uinhsBfqro3sxeF5KUn9V/5j7Dj47l4FSgKfOb25+9U1VuDFrSf+Hgu8gQfz8VsoLWIbACSgIdV9WDwovYPH8/FMOA9EXkQ51JLn0j8YikiH+Ncaizp9sc8DRQAUNWxOP0zbYGtwD9AX5+OG4HnyhhjTC4K1UtPxhhjQoQlCmOMMRmyRGGMMSZDliiMMcZkyBKFMcaYDFmiMAElIiVEJN792Scie7zuF/Tze7cSka+ysP+VInLCK754EYnKreOnc4z8IpLkvtc6t0Jy4Wwc530RuVpE8nlXUxWRKBFZlJMYTd5jicIElKoeVNVoVY0GxgJveO67Bd1CzSav+KJVNSkA73nUPT813fsDsnoAVe2rqptw/o+P8Ho8SVWvz50wTV5hicKEDBGZISKr3LUTYt3HrhCRLSJysftteImI3JDe/mkcs527TsFivCqKikhREflARJa7axTckoU4G4rIj+7rfhCRc8piiMgNIrLabRn8JCJF3G/3r7sthbUi0jmj93EnhC0CrnSP+Yj72nUicp/7WDERmeW+1zrPMUVksYhEw/+3d3YhVlVRHP/9NXUcRkcFtRCEmAFRoXzRFD8IUhF9qB4kJRQffRIMMUIQlD40LcKgHkTKz3oZiVAaQxDSQXNAdAY1RB98EHqKEU2RHpYPa109DmeO1x6cy8z6webue/Y+ey8ul73P2uve/2IXMC7sOBQeS1/06ZC0vGDzEUnvShor6WDYeEnSkno/m2SIMtj66VmGb8EVPbcU3k+K12bgGk/zKGzEpaM/wWWzK/sX2ptxbZs2XAytA/gl2r4E1kR9InADaOp3fzvwELgcZV9cbwVGRn0Fkf8DWFoY/zfgrai34P8Y/gDojPqruIrnlH5zvgL0RX0UcAL3KObheRaacU2r68AbMeb3hftb4/UcMKc4Xsn4q4EDUW8Ke8YAHwP74/ps4DYwerC/L1kGr6RHkTQSmyVdAc7jYmVt8ER6YDIuN7D1ef0LzAJumNktMzPgaKFtObBN0mXgDL5QTi+xqXj0tCmuTQCOy7OI7cUX0/50Ad/Ek/948yOrRcAx8+Ofv/HFvEywcFzY1Y0n5PkRWAx0mNkDM7uH6zgtAnqAFZJ2SVpoZndLxhuIk8AySaOAVbhe2qMY9zCAmV3FRePaX2DcZIjRkFpPyfBDrn67BJhvZg/jqKgp2lqA1/An8Rbg36r+/RhIo0bAe2Z263+Y+xlwysy+k9SOewnPTmr2qaRf8QW4W9LblEs8l1GLUTw1VuVJuczsulwddyWwR9IJM/u8nknM7IGkLmAZ7pn8UJuuTjuTYUJ6FEmj0Ar8E4v+bGBuoW0P/lS9E1fJfV7/GtfwxD2vx0K7ttB2Cqh5CDUF3hex9U7UN5R1kNRmZj1m9gWejnQGnp51TcRapgILgXqVbf8A3o/4QQsebzkraRpw38wOA1/jaTCfYJ7UqybFX8bPuAT3AuB0Ya4P476Z+CZ9s047kyFIbhRJo3ASaI6jpO3AnwCS3gHeBL4ys4PACEnrBupfxDw97EY8XnCWZyW2d8T9vZKu4vGSetmNP713VfTZEsHlHqAP+B1P2fsXHms4DXxkdcrCm9lF4Cf8OOoCHpfoxT+b7jiq2gqUeRMHgB5Jh0raOnH57U4z+y+ufQuMldSLH9ett8b8RVrykkj12CRJkqSS9CiSJEmSSnKjSJIkSSrJjSJJkiSpJDeKJEmSpJLcKJIkSZJKcqNIkiRJKsmNIkmSJKnkMTQSo5b00TpMAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "[mean: 0.50000, std: 0.00000, params: {'subsample': 0.1},\n",
       " mean: 0.50000, std: 0.00000, params: {'subsample': 0.2},\n",
       " mean: 0.72558, std: 0.01630, params: {'subsample': 0.30000000000000004},\n",
       " mean: 0.72811, std: 0.01605, params: {'subsample': 0.4},\n",
       " mean: 0.73050, std: 0.01713, params: {'subsample': 0.5},\n",
       " mean: 0.73042, std: 0.01220, params: {'subsample': 0.6},\n",
       " mean: 0.72915, std: 0.01467, params: {'subsample': 0.7000000000000001},\n",
       " mean: 0.72981, std: 0.01250, params: {'subsample': 0.8},\n",
       " mean: 0.73121, std: 0.01269, params: {'subsample': 0.9}]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'subsample': 0.9}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "0.7312113052217616"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "t0 = time()\n",
    "param7 = {'subsample': np.arange(0.1, 1, 0.1)} \n",
    "grid7 = GridSearchCV(estimator = GradientBoostingClassifier(learning_rate = 0.1,\n",
    "                                                            n_estimators = 90,\n",
    "                                                            max_depth=3,\n",
    "                                                            max_features = 24,\n",
    "                                                            subsample = 0.8,\n",
    "                                                            min_samples_leaf= 65,\n",
    "                                                            min_samples_split = 1800,\n",
    "                                                            random_state = 10), \n",
    "param_grid = param7, scoring = 'roc_auc', n_jobs = 4, iid = False, cv = 5) \n",
    "grid7.fit(x_train, y_train) \n",
    "\n",
    "print(\"Tempo de execução:\", round(time()-t0, 3), \"s\")\n",
    "draw_roc_score(grid7, x_train, y_train, x_test, y_test)\n",
    "display(grid7.grid_scores_, grid7.best_params_, grid7.best_score_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "O valor ideal para subsample é 0.9."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visualizando Score Roc nos dados de teste do estimador do passo 7"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "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"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Score Roc: 74.0%\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import roc_curve\n",
    "from sklearn.metrics import roc_auc_score\n",
    "\n",
    "clf_7 = grid7.best_estimator_\n",
    "\n",
    "y_probas = clf_7.predict_proba(x_test)\n",
    "y_scores = y_probas[:, 1] \n",
    "\n",
    "fpr1, tpr1, thresholds = roc_curve(y_test, y_scores)\n",
    "\n",
    "plt.plot(fpr1, tpr1, linewidth=1.5, label=None)\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.axis([0, 1, 0, 1])\n",
    "plt.xlabel('Taxa de Falso Positivo')\n",
    "plt.ylabel('Taxa de Verdadeiro Positivo')\n",
    "plt.show()\n",
    "\n",
    "print(\"Score Roc: {}%\".format(np.round(roc_auc_score(y_test, y_scores)*100), 2))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Análise Final do tunning\n",
    "\n",
    "Como podemos perceber, encontrei o maior **Score Roc** nos dados de teste(74%) no passo 5 e no melhor estimador gerado no passo 7, mas o passo 6 foi o que apresentou a menor diferença entre os scores de treino e teste. \n",
    "Logo abaixo segue a configuração do estimador 5, 6 e 7:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GradientBoostingClassifier(criterion='friedman_mse', init=None,\n",
       "              learning_rate=0.1, loss='deviance', max_depth=3,\n",
       "              max_features=24, max_leaf_nodes=None,\n",
       "              min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "              min_samples_leaf=10, min_samples_split=1700,\n",
       "              min_weight_fraction_leaf=0.0, n_estimators=90,\n",
       "              presort='auto', random_state=10, subsample=0.8, verbose=0,\n",
       "              warm_start=False)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "GradientBoostingClassifier(criterion='friedman_mse', init=None,\n",
       "              learning_rate=0.1, loss='deviance', max_depth=3,\n",
       "              max_features=24, max_leaf_nodes=None,\n",
       "              min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "              min_samples_leaf=65, min_samples_split=1800,\n",
       "              min_weight_fraction_leaf=0.0, n_estimators=90,\n",
       "              presort='auto', random_state=10, subsample=0.8, verbose=0,\n",
       "              warm_start=False)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "GradientBoostingClassifier(criterion='friedman_mse', init=None,\n",
       "              learning_rate=0.1, loss='deviance', max_depth=3,\n",
       "              max_features=24, max_leaf_nodes=None,\n",
       "              min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "              min_samples_leaf=65, min_samples_split=1800,\n",
       "              min_weight_fraction_leaf=0.0, n_estimators=90,\n",
       "              presort='auto', random_state=10, subsample=0.9, verbose=0,\n",
       "              warm_start=False)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display(grid5.best_estimator_)\n",
    "display(grid6.best_estimator_)\n",
    "display(clf_7)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "O primeiro ponto a ser notado é as diferenças nos hiperparâmetros min_samples_leaf, min_samples_split e subsample, onde, para o min_samples_leaf e min_samples_split, o ideal é que se escolham os valores baixos para controlar o over-fitting, já para subsample, o ideal é que o valor seja menor que 1 com a finalidade de reduzir a variação. O segundo ponto é a diferença entre o score roc de treino e teste, onde a menor diferença representa que o modelo está generalizando bem para dados não vistos. Dessa maneira, levando em consideração os pontos citados anteriormente, os modelos encontrados nos **passo 5 e 6** apresentam os melhores ajustes. O mode do passo 5 apresenta os melhores ajustes definidos no primeiro ponto e o 6 apresenta a menor diferença entre os scores, dessa forma, irei ficar o passo 6 por generalizar melhor."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GradientBoostingClassifier(criterion='friedman_mse', init=None,\n",
       "              learning_rate=0.1, loss='deviance', max_depth=3,\n",
       "              max_features=24, max_leaf_nodes=None,\n",
       "              min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "              min_samples_leaf=65, min_samples_split=1800,\n",
       "              min_weight_fraction_leaf=0.0, n_estimators=90,\n",
       "              presort='auto', random_state=10, subsample=0.8, verbose=0,\n",
       "              warm_start=False)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf = GradientBoostingClassifier(criterion='friedman_mse', init=None,\n",
    "              learning_rate=0.1, loss='deviance', max_depth=3,\n",
    "              max_features=24, max_leaf_nodes=None,\n",
    "              min_impurity_decrease=0.0, min_impurity_split=None,\n",
    "              min_samples_leaf=65, min_samples_split=1800,\n",
    "              min_weight_fraction_leaf=0.0, n_estimators=90,\n",
    "              presort='auto', random_state=10, subsample=0.8, verbose=0,\n",
    "              warm_start=False)\n",
    "\n",
    "clf.fit(x_train,y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizando o desempenho do classificador selecionado"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 396x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "clf.predict(x_train)\n",
    "y_pred = clf.predict(x_test)\n",
    "\n",
    "cm = confusion_matrix(y_test, y_pred) \n",
    "\n",
    "cm_df = pd.DataFrame(cm,\n",
    "                     index = ['Bom Pagador','Inadimplente'], \n",
    "                     columns = ['Bom Pagador','Inadimplente'])\n",
    "\n",
    "plt.figure(figsize=(5.5,4))\n",
    "sns.heatmap(cm_df, annot=True, fmt='g')\n",
    "plt.title('GradientBoostingClassifier \\nAccuracy:{0:.3f}%'.format(accuracy_score(y_test, y_pred)* 100))\n",
    "plt.ylabel('True label')\n",
    "plt.xlabel('Predicted label')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Como podemos perceber na matriz de confusão, o classificador possui uma precisão de 91,8% na identificação da situação dos mutuários."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}