incendios_alberta_pre_processamento_versao_final.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "22407056",
   "metadata": {},
   "source": [
    "# Uma abordagem inovadora utilizando aprendizagem de máquina integrada ao Azure para classificação da causa de incêndios florestais em Alberta no Canadá"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bDDJRMD6vH0D",
   "metadata": {
    "id": "bDDJRMD6vH0D"
   },
   "source": [
    "## Importação das Bibliotecas Necessárias"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "469ea58f",
   "metadata": {
    "id": "469ea58f"
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import datetime as dt\n",
    "from datetime import datetime, date, timezone\n",
    "from collections import Counter\n",
    "import seaborn as sn\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "VFVPCqnGvUl6",
   "metadata": {
    "id": "VFVPCqnGvUl6"
   },
   "source": [
    "## Pré-processamento da Dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "CA9RZvuuveNy",
   "metadata": {
    "id": "CA9RZvuuveNy"
   },
   "source": [
    "### Importação do Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "UKhaRVckaigL",
   "metadata": {
    "id": "UKhaRVckaigL"
   },
   "outputs": [],
   "source": [
    "# !pip freeze > requirements.txt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "SPT8cMNAlX73",
   "metadata": {
    "id": "SPT8cMNAlX73"
   },
   "outputs": [],
   "source": [
    "# pd.set_option('display.max_columns', None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "66RWQa5KAC3f",
   "metadata": {
    "id": "66RWQa5KAC3f"
   },
   "outputs": [],
   "source": [
    "df_96_05 = pd.read_csv(\"alberta_fires_1996to2005.csv\", encoding='cp1252')\n",
    "df_06_18 = pd.read_csv(\"alberta_fires_2006to2018.csv\", encoding='cp1252')\n",
    "df_96_18 = pd.concat([df_96_05, df_06_18])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "J1e9qwW95DrE",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "J1e9qwW95DrE",
    "outputId": "9c990fa8-d24d-4934-c0bf-b4fceb1a479e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['fire_number', 'fire_name', 'fire_year', 'calendar_year',\n",
       "       'assessment_datetime', 'assessment_hectares', 'current_size',\n",
       "       'size_class', 'fire_location_latitude', 'fire_location_longitude',\n",
       "       'fire_origin', 'general_cause_desc', 'industry_identifier_desc',\n",
       "       'responsible_group_desc', 'activity_class', 'true_cause',\n",
       "       'permit_detail_desc', 'fire_start_date', 'det_agent_type', 'det_agent',\n",
       "       'discovered_date', 'reported_date', 'start_for_fire_date',\n",
       "       'fire_fighting_start_date', 'fire_fighting_start_size',\n",
       "       'initial_action_by', 'fire_type', 'fire_position_on_slope',\n",
       "       'weather_conditions_over_fire', 'fuel_type', 'other_fuel_type',\n",
       "       'bh_fs_date', 'bh_hectares', 'uc_fs_date', 'uc_hectares', 'to_fs_date',\n",
       "       'to_hectares', 'ex_fs_date', 'ex_hectares'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_96_18.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "12210d34",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "12210d34",
    "outputId": "a83a6560-0781-42ec-c35c-72a46c402aaa"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(31196, 39)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_96_18.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "708c9511",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "708c9511",
    "outputId": "88874aa6-9036-40fe-a920-e4773d276591"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-56d9bca9-7c3f-4f46-8819-dc014388d727\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <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>9888</th>\n",
       "      <th>16693</th>\n",
       "      <th>356</th>\n",
       "      <th>7420</th>\n",
       "      <th>8902</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>fire_number</th>\n",
       "      <td>CWF106</td>\n",
       "      <td>LWF095</td>\n",
       "      <td>EWF123</td>\n",
       "      <td>GWF052</td>\n",
       "      <td>HWF319</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fire_name</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fire_year</th>\n",
       "      <td>2012</td>\n",
       "      <td>2016</td>\n",
       "      <td>2006</td>\n",
       "      <td>2003</td>\n",
       "      <td>2004</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>calendar_year</th>\n",
       "      <td>2012</td>\n",
       "      <td>2016</td>\n",
       "      <td>2006</td>\n",
       "      <td>2003</td>\n",
       "      <td>2004</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>assessment_datetime</th>\n",
       "      <td>2012-05-21 15:55:00</td>\n",
       "      <td>2016-06-26 20:25:00</td>\n",
       "      <td>2006-07-01 17:50:00</td>\n",
       "      <td>2003-07-24 14:10:00</td>\n",
       "      <td>2004-08-01 18:03:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>assessment_hectares</th>\n",
       "      <td>0.01</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.03</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>current_size</th>\n",
       "      <td>0.01</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.11</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>size_class</th>\n",
       "      <td>A</td>\n",
       "      <td>A</td>\n",
       "      <td>B</td>\n",
       "      <td>A</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fire_location_latitude</th>\n",
       "      <td>49.815383</td>\n",
       "      <td>56.103483</td>\n",
       "      <td>53.93425</td>\n",
       "      <td>55.8724</td>\n",
       "      <td>59.382751</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fire_location_longitude</th>\n",
       "      <td>-113.957033</td>\n",
       "      <td>-112.605833</td>\n",
       "      <td>-118.9567</td>\n",
       "      <td>-119.6244</td>\n",
       "      <td>-119.616232</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fire_origin</th>\n",
       "      <td>Provincial Land</td>\n",
       "      <td>Provincial Land</td>\n",
       "      <td>Private Land</td>\n",
       "      <td>Provincial Land</td>\n",
       "      <td>Provincial Land</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>general_cause_desc</th>\n",
       "      <td>Recreation</td>\n",
       "      <td>Lightning</td>\n",
       "      <td>Railroad</td>\n",
       "      <td>Lightning</td>\n",
       "      <td>Lightning</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>industry_identifier_desc</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>responsible_group_desc</th>\n",
       "      <td>Campers</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Operators</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>activity_class</th>\n",
       "      <td>Cooking and Warming</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Rail Transportation</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>true_cause</th>\n",
       "      <td>Unsafe Fire</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Unpredictable Event</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>permit_detail_desc</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fire_start_date</th>\n",
       "      <td>2012-05-21 12:00:00</td>\n",
       "      <td>2016-06-26 16:30:00</td>\n",
       "      <td>2006-07-01 17:30:00</td>\n",
       "      <td>2003-07-23 22:00:00</td>\n",
       "      <td>2004-07-27 16:40:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>det_agent_type</th>\n",
       "      <td>GRP</td>\n",
       "      <td>UNP</td>\n",
       "      <td>UNP</td>\n",
       "      <td>UNP</td>\n",
       "      <td>LKT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>det_agent</th>\n",
       "      <td>PATR</td>\n",
       "      <td>LFS</td>\n",
       "      <td>310</td>\n",
       "      <td>310</td>\n",
       "      <td>PD</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>discovered_date</th>\n",
       "      <td>2012-05-21 15:55:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2003-07-24 13:00:00</td>\n",
       "      <td>2004-08-01 16:53:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>reported_date</th>\n",
       "      <td>2012-05-21 15:55:00</td>\n",
       "      <td>2016-06-26 20:25:00</td>\n",
       "      <td>2006-07-01 17:39:00</td>\n",
       "      <td>2003-07-24 13:00:00</td>\n",
       "      <td>2004-08-01 16:57:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>start_for_fire_date</th>\n",
       "      <td>2012-05-21 15:55:00</td>\n",
       "      <td>2016-06-26 20:25:00</td>\n",
       "      <td>2006-07-01 17:41:00</td>\n",
       "      <td>2003-07-24 13:00:00</td>\n",
       "      <td>2004-08-01 17:29:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fire_fighting_start_date</th>\n",
       "      <td>2012-05-21 15:55:00</td>\n",
       "      <td>2016-06-27 10:30:00</td>\n",
       "      <td>2006-07-01 17:50:00</td>\n",
       "      <td>2003-07-24 14:10:00</td>\n",
       "      <td>2004-08-01 18:11:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fire_fighting_start_size</th>\n",
       "      <td>0.01</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.03</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>initial_action_by</th>\n",
       "      <td>FPD Staff</td>\n",
       "      <td>RAP</td>\n",
       "      <td>HAC1H</td>\n",
       "      <td>HAC1H</td>\n",
       "      <td>Air Tanker</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fire_type</th>\n",
       "      <td>NaN</td>\n",
       "      <td>Crown</td>\n",
       "      <td>Surface</td>\n",
       "      <td>Surface</td>\n",
       "      <td>Ground</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fire_position_on_slope</th>\n",
       "      <td>NaN</td>\n",
       "      <td>Flat</td>\n",
       "      <td>Upper 1/3</td>\n",
       "      <td>Flat</td>\n",
       "      <td>Flat</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>weather_conditions_over_fire</th>\n",
       "      <td>NaN</td>\n",
       "      <td>Clear</td>\n",
       "      <td>Clear</td>\n",
       "      <td>Cloudy</td>\n",
       "      <td>CB Dry</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fuel_type</th>\n",
       "      <td>NaN</td>\n",
       "      <td>M2</td>\n",
       "      <td>O1a</td>\n",
       "      <td>M2</td>\n",
       "      <td>C2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>other_fuel_type</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bh_fs_date</th>\n",
       "      <td>2012-05-21 15:55:00</td>\n",
       "      <td>2016-06-26 20:27:00</td>\n",
       "      <td>2006-07-01 20:12:00</td>\n",
       "      <td>2003-07-24 14:30:00</td>\n",
       "      <td>2004-08-01 18:03:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bh_hectares</th>\n",
       "      <td>0.01</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.03</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>uc_fs_date</th>\n",
       "      <td>2012-05-21 15:55:00</td>\n",
       "      <td>2016-06-27 10:10:00</td>\n",
       "      <td>2006-07-01 20:20:00</td>\n",
       "      <td>2003-07-24 15:00:00</td>\n",
       "      <td>2004-08-01 21:00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>uc_hectares</th>\n",
       "      <td>0.01</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.03</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>to_fs_date</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>to_hectares</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ex_fs_date</th>\n",
       "      <td>2012-05-21 16:10:00</td>\n",
       "      <td>2016-06-27 10:16:00</td>\n",
       "      <td>2006-07-01 20:44:00</td>\n",
       "      <td>2003-07-25 10:00:00</td>\n",
       "      <td>2004-08-01 21:37:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ex_hectares</th>\n",
       "      <td>0.01</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.11</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-56d9bca9-7c3f-4f46-8819-dc014388d727')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-56d9bca9-7c3f-4f46-8819-dc014388d727 button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-56d9bca9-7c3f-4f46-8819-dc014388d727');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "                                            9888                 16693  \\\n",
       "fire_number                                CWF106               LWF095   \n",
       "fire_name                                     NaN                  NaN   \n",
       "fire_year                                    2012                 2016   \n",
       "calendar_year                                2012                 2016   \n",
       "assessment_datetime           2012-05-21 15:55:00  2016-06-26 20:25:00   \n",
       "assessment_hectares                          0.01                 0.01   \n",
       "current_size                                 0.01                 0.01   \n",
       "size_class                                      A                    A   \n",
       "fire_location_latitude                  49.815383            56.103483   \n",
       "fire_location_longitude               -113.957033          -112.605833   \n",
       "fire_origin                       Provincial Land      Provincial Land   \n",
       "general_cause_desc                     Recreation            Lightning   \n",
       "industry_identifier_desc                      NaN                  NaN   \n",
       "responsible_group_desc                    Campers                  NaN   \n",
       "activity_class                Cooking and Warming                  NaN   \n",
       "true_cause                            Unsafe Fire                  NaN   \n",
       "permit_detail_desc                            NaN                  NaN   \n",
       "fire_start_date               2012-05-21 12:00:00  2016-06-26 16:30:00   \n",
       "det_agent_type                                GRP                  UNP   \n",
       "det_agent                                    PATR                  LFS   \n",
       "discovered_date               2012-05-21 15:55:00                  NaN   \n",
       "reported_date                 2012-05-21 15:55:00  2016-06-26 20:25:00   \n",
       "start_for_fire_date           2012-05-21 15:55:00  2016-06-26 20:25:00   \n",
       "fire_fighting_start_date      2012-05-21 15:55:00  2016-06-27 10:30:00   \n",
       "fire_fighting_start_size                     0.01                 0.01   \n",
       "initial_action_by                       FPD Staff                  RAP   \n",
       "fire_type                                     NaN                Crown   \n",
       "fire_position_on_slope                        NaN                 Flat   \n",
       "weather_conditions_over_fire                  NaN                Clear   \n",
       "fuel_type                                     NaN                   M2   \n",
       "other_fuel_type                               NaN                  NaN   \n",
       "bh_fs_date                    2012-05-21 15:55:00  2016-06-26 20:27:00   \n",
       "bh_hectares                                  0.01                 0.01   \n",
       "uc_fs_date                    2012-05-21 15:55:00  2016-06-27 10:10:00   \n",
       "uc_hectares                                  0.01                 0.01   \n",
       "to_fs_date                                    NaN                  NaN   \n",
       "to_hectares                                   NaN                  NaN   \n",
       "ex_fs_date                    2012-05-21 16:10:00  2016-06-27 10:16:00   \n",
       "ex_hectares                                  0.01                 0.01   \n",
       "\n",
       "                                            356                  7420   \\\n",
       "fire_number                                EWF123               GWF052   \n",
       "fire_name                                     NaN                  NaN   \n",
       "fire_year                                    2006                 2003   \n",
       "calendar_year                                2006                 2003   \n",
       "assessment_datetime           2006-07-01 17:50:00  2003-07-24 14:10:00   \n",
       "assessment_hectares                          0.03                 0.01   \n",
       "current_size                                 0.11                 0.01   \n",
       "size_class                                      B                    A   \n",
       "fire_location_latitude                   53.93425              55.8724   \n",
       "fire_location_longitude                 -118.9567            -119.6244   \n",
       "fire_origin                          Private Land      Provincial Land   \n",
       "general_cause_desc                       Railroad            Lightning   \n",
       "industry_identifier_desc                      NaN                  NaN   \n",
       "responsible_group_desc                  Operators                  NaN   \n",
       "activity_class                Rail Transportation                  NaN   \n",
       "true_cause                    Unpredictable Event                  NaN   \n",
       "permit_detail_desc                            NaN                  NaN   \n",
       "fire_start_date               2006-07-01 17:30:00  2003-07-23 22:00:00   \n",
       "det_agent_type                                UNP                  UNP   \n",
       "det_agent                                     310                  310   \n",
       "discovered_date                               NaN  2003-07-24 13:00:00   \n",
       "reported_date                 2006-07-01 17:39:00  2003-07-24 13:00:00   \n",
       "start_for_fire_date           2006-07-01 17:41:00  2003-07-24 13:00:00   \n",
       "fire_fighting_start_date      2006-07-01 17:50:00  2003-07-24 14:10:00   \n",
       "fire_fighting_start_size                     0.03                 0.01   \n",
       "initial_action_by                           HAC1H                HAC1H   \n",
       "fire_type                                 Surface              Surface   \n",
       "fire_position_on_slope                  Upper 1/3                 Flat   \n",
       "weather_conditions_over_fire                Clear               Cloudy   \n",
       "fuel_type                                     O1a                   M2   \n",
       "other_fuel_type                               NaN                  NaN   \n",
       "bh_fs_date                    2006-07-01 20:12:00  2003-07-24 14:30:00   \n",
       "bh_hectares                                  0.03                 0.01   \n",
       "uc_fs_date                    2006-07-01 20:20:00  2003-07-24 15:00:00   \n",
       "uc_hectares                                  0.03                 0.01   \n",
       "to_fs_date                                    NaN                  NaN   \n",
       "to_hectares                                   NaN                  NaN   \n",
       "ex_fs_date                    2006-07-01 20:44:00  2003-07-25 10:00:00   \n",
       "ex_hectares                                  0.11                 0.01   \n",
       "\n",
       "                                            8902   \n",
       "fire_number                                HWF319  \n",
       "fire_name                                     NaN  \n",
       "fire_year                                    2004  \n",
       "calendar_year                                2004  \n",
       "assessment_datetime           2004-08-01 18:03:00  \n",
       "assessment_hectares                           0.5  \n",
       "current_size                                  0.3  \n",
       "size_class                                      B  \n",
       "fire_location_latitude                  59.382751  \n",
       "fire_location_longitude               -119.616232  \n",
       "fire_origin                       Provincial Land  \n",
       "general_cause_desc                      Lightning  \n",
       "industry_identifier_desc                      NaN  \n",
       "responsible_group_desc                        NaN  \n",
       "activity_class                                NaN  \n",
       "true_cause                                    NaN  \n",
       "permit_detail_desc                            NaN  \n",
       "fire_start_date               2004-07-27 16:40:00  \n",
       "det_agent_type                                LKT  \n",
       "det_agent                                      PD  \n",
       "discovered_date               2004-08-01 16:53:00  \n",
       "reported_date                 2004-08-01 16:57:00  \n",
       "start_for_fire_date           2004-08-01 17:29:00  \n",
       "fire_fighting_start_date      2004-08-01 18:11:00  \n",
       "fire_fighting_start_size                      0.3  \n",
       "initial_action_by                      Air Tanker  \n",
       "fire_type                                  Ground  \n",
       "fire_position_on_slope                       Flat  \n",
       "weather_conditions_over_fire               CB Dry  \n",
       "fuel_type                                      C2  \n",
       "other_fuel_type                               NaN  \n",
       "bh_fs_date                    2004-08-01 18:03:00  \n",
       "bh_hectares                                   0.5  \n",
       "uc_fs_date                    2004-08-01 21:00:00  \n",
       "uc_hectares                                   0.3  \n",
       "to_fs_date                                    NaN  \n",
       "to_hectares                                   NaN  \n",
       "ex_fs_date                    2004-08-01 21:37:00  \n",
       "ex_hectares                                   0.3  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_96_18.sample(5).T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "wFtQdaaVXjdk",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "wFtQdaaVXjdk",
    "outputId": "139570fc-729d-49bb-a15f-12e81c2c0a28"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 31196 entries, 0 to 19843\n",
      "Data columns (total 39 columns):\n",
      " #   Column                        Non-Null Count  Dtype  \n",
      "---  ------                        --------------  -----  \n",
      " 0   fire_number                   31196 non-null  object \n",
      " 1   fire_name                     637 non-null    object \n",
      " 2   fire_year                     31196 non-null  int64  \n",
      " 3   calendar_year                 31196 non-null  int64  \n",
      " 4   assessment_datetime           31196 non-null  object \n",
      " 5   assessment_hectares           31195 non-null  float64\n",
      " 6   current_size                  31195 non-null  float64\n",
      " 7   size_class                    31196 non-null  object \n",
      " 8   fire_location_latitude        31196 non-null  float64\n",
      " 9   fire_location_longitude       31196 non-null  float64\n",
      " 10  fire_origin                   31186 non-null  object \n",
      " 11  general_cause_desc            31196 non-null  object \n",
      " 12  industry_identifier_desc      978 non-null    object \n",
      " 13  responsible_group_desc        13257 non-null  object \n",
      " 14  activity_class                17217 non-null  object \n",
      " 15  true_cause                    15949 non-null  object \n",
      " 16  permit_detail_desc            1078 non-null   object \n",
      " 17  fire_start_date               30647 non-null  object \n",
      " 18  det_agent_type                31194 non-null  object \n",
      " 19  det_agent                     31194 non-null  object \n",
      " 20  discovered_date               28498 non-null  object \n",
      " 21  reported_date                 31194 non-null  object \n",
      " 22  start_for_fire_date           29536 non-null  object \n",
      " 23  fire_fighting_start_date      24224 non-null  object \n",
      " 24  fire_fighting_start_size      24242 non-null  float64\n",
      " 25  initial_action_by             31195 non-null  object \n",
      " 26  fire_type                     28943 non-null  object \n",
      " 27  fire_position_on_slope        28841 non-null  object \n",
      " 28  weather_conditions_over_fire  28844 non-null  object \n",
      " 29  fuel_type                     24322 non-null  object \n",
      " 30  other_fuel_type               6218 non-null   object \n",
      " 31  bh_fs_date                    31192 non-null  object \n",
      " 32  bh_hectares                   31192 non-null  float64\n",
      " 33  uc_fs_date                    31191 non-null  object \n",
      " 34  uc_hectares                   31191 non-null  float64\n",
      " 35  to_fs_date                    3156 non-null   object \n",
      " 36  to_hectares                   3156 non-null   float64\n",
      " 37  ex_fs_date                    31195 non-null  object \n",
      " 38  ex_hectares                   31195 non-null  float64\n",
      "dtypes: float64(9), int64(2), object(28)\n",
      "memory usage: 9.5+ MB\n"
     ]
    }
   ],
   "source": [
    "df_96_18.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "O19rIMI108wY",
   "metadata": {
    "id": "O19rIMI108wY"
   },
   "source": [
    "### Avaliação de atributos para tarefa de classificação"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bNFEGVdvy8Fi",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "bNFEGVdvy8Fi",
    "outputId": "55c4c9ee-d2be-4ddb-9908-aa01eb54005d"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "787"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_96_18[df_96_18['general_cause_desc'] == 'Undetermined']['general_cause_desc'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4waG3-GJzf2Q",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "4waG3-GJzf2Q",
    "outputId": "acdd2596-c025-45f5-fb07-a6c3d9b18a77"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Lightning               13080\n",
       "Recreation               6175\n",
       "Resident                 5392\n",
       "Incendiary               1899\n",
       "Power Line Industry       988\n",
       "Other Industry            891\n",
       "Undetermined              787\n",
       "Forest Industry           514\n",
       "Oil & Gas Industry        470\n",
       "Agriculture Industry      341\n",
       "Railroad                  245\n",
       "Miscellaneous Known       225\n",
       "Government                 90\n",
       "Restart                    74\n",
       "Prescribed Fire            24\n",
       "Under Investigation         1\n",
       "Name: general_cause_desc, dtype: int64"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_96_18['general_cause_desc'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "V2eOgI2GCvOm",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "V2eOgI2GCvOm",
    "outputId": "3d633bd3-31cd-4961-ddc1-70dd00beb94f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "A    19386\n",
       "B     9185\n",
       "C     1663\n",
       "E      520\n",
       "D      442\n",
       "Name: size_class, dtype: int64"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_96_18['size_class'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "no8Pqepk1K7H",
   "metadata": {
    "id": "no8Pqepk1K7H"
   },
   "source": [
    "### Correção de alguns registros dos atributos 'fire type' and 'fuel type'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "HWsGliv8vRyI",
   "metadata": {
    "id": "HWsGliv8vRyI"
   },
   "outputs": [],
   "source": [
    "df_96_18['fire_type'] = df_96_18['fire_type'].map(lambda v: v.replace('   Surface', 'Surface'), na_action = 'ignore')\n",
    "df_96_18['fire_type'] = df_96_18['fire_type'].map(lambda v: v.replace('  Surface', 'Surface'), na_action = 'ignore')\n",
    "df_96_18['fire_type'] = df_96_18['fire_type'].map(lambda v: v.replace('    Ground', 'Ground'), na_action = 'ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "E6Hpc-txv-J6",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "E6Hpc-txv-J6",
    "outputId": "f49afd9e-447c-470a-b01c-32b53454c497"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[9955]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "void_reg_fire_type_index = list(df_96_18[df_96_18['fire_type'] == '  ']['fire_type'].index)\n",
    "void_reg_fire_type_index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "-vxSKAn2wMvk",
   "metadata": {
    "id": "-vxSKAn2wMvk"
   },
   "outputs": [],
   "source": [
    "# Remoção dos registros nos quais \"fuel_type\" está com espaço em branco\n",
    "df_96_18.drop(void_reg_fire_type_index, axis=0, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "gBsSDWc2z_hf",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "gBsSDWc2z_hf",
    "outputId": "45cee9de-9c31-4ba8-a194-259414e8ebc3"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Surface    19947\n",
       "Ground      6678\n",
       "Crown       2315\n",
       "               1\n",
       "Name: fire_type, dtype: int64"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_96_18['fire_type'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1ZCSA8iAr26v",
   "metadata": {
    "id": "1ZCSA8iAr26v"
   },
   "outputs": [],
   "source": [
    "# Correção de um registro no qual \"C2\" foi armazenado como \" C2\"\n",
    "df_96_18['fuel_type'] = df_96_18['fuel_type'].map(lambda v: v.replace(' C2', 'C2'), na_action = 'ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "DXuly5h9s4bc",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "DXuly5h9s4bc",
    "outputId": "e4961eca-40f5-4ec5-c590-e4f666e791bc"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[3462, 10959, 11089, 11210]"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Busca dos indices para os quais o registro de \"fuel_type\" é igual a espaço vazio\n",
    "void_reg_fuel_type_index = list(df_96_18[df_96_18['fuel_type'] == ' ']['fuel_type'].index)\n",
    "void_reg_fuel_type_index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0ycFMW4Ntdxj",
   "metadata": {
    "id": "0ycFMW4Ntdxj"
   },
   "outputs": [],
   "source": [
    "# Remoção dos registros nos quais \"fuel_type\" está com espaço em branco\n",
    "df_96_18.drop(void_reg_fuel_type_index, axis=0, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3QkhpZN5aeBX",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "3QkhpZN5aeBX",
    "outputId": "45bdd9c1-9068-4a4c-8cd6-ffefa26db8e7"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "C2     9582\n",
       "O1a    4184\n",
       "M2     2980\n",
       "O1b    2301\n",
       "C3     1081\n",
       "C1     1030\n",
       "M1      901\n",
       "D1      720\n",
       "S2      652\n",
       "S1      606\n",
       "C4      225\n",
       "C7       24\n",
       "M3       16\n",
       "M4        7\n",
       "C6        3\n",
       "S3        3\n",
       "Name: fuel_type, dtype: int64"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_96_18['fuel_type'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "rVq35B0jwc93",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "rVq35B0jwc93",
    "outputId": "61ef716c-8596-4b0a-bca7-1bac04f5b425"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(31186, 39)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Após exclusão de registros, reordena os index\n",
    "df_96_18.reset_index(drop=True, inplace=True)\n",
    "df_96_18.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "BUUOyRC82H-r",
   "metadata": {
    "id": "BUUOyRC82H-r"
   },
   "source": [
    "### Verificação da existência de registros duplicados"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "oDB2Nsf227Hb",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "oDB2Nsf227Hb",
    "outputId": "fe3ccd66-dcf0-40b5-909c-6d22eca91489"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "False    31186\n",
       "dtype: int64"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_96_18.duplicated(keep='first').value_counts()\n",
    "# 'False' corresponde aos dados não duplicados"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "U3G4g_qX2SFO",
   "metadata": {
    "id": "U3G4g_qX2SFO"
   },
   "source": [
    "### Preenchimento dos NAN do atributo `fuel_type`"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ZHGhOAa72mlY",
   "metadata": {
    "id": "ZHGhOAa72mlY"
   },
   "source": [
    "* Quando a classificação do tipo de combustível (fuel_type) é \"outro tipo de combustível\" (other_fuel_type), o atributo `fuel_type` não é preenchido e o registro é feito no atributo `other_fuel_type`;\n",
    "* Portanto, foram verificados os registros NAN para `fuel_type` que correspondiam a registros preenchidos em `other_fuel_type` e substituiu-se tais registros por uma classe adicional chamada `other_fuel` (OF)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "-y2vtnzdxZRO",
   "metadata": {
    "id": "-y2vtnzdxZRO"
   },
   "outputs": [],
   "source": [
    "for i,fuel in enumerate(df_96_18['fuel_type']):\n",
    "  if pd.isna(fuel):\n",
    "    if not pd.isna(df_96_18.loc[i,'other_fuel_type']):\n",
    "      df_96_18.loc[i,'fuel_type'] = 'OF'               # OF = Other Fuel"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "T2qh2Rf3zr69",
   "metadata": {
    "id": "T2qh2Rf3zr69"
   },
   "outputs": [],
   "source": [
    "df_96_18['fuel_type'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eWUtO-glvj0U",
   "metadata": {
    "id": "eWUtO-glvj0U"
   },
   "source": [
    "### Verificação do Percentual de Dados Ausentes em cada Coluna do Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "vicwZaMwkGn5",
   "metadata": {
    "id": "vicwZaMwkGn5"
   },
   "outputs": [],
   "source": [
    "(df_96_18.isna().sum()/df_96_18.shape[0])*100"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "lVnAQCev4zuo",
   "metadata": {
    "id": "lVnAQCev4zuo"
   },
   "source": [
    "### Descrição de alguns atributos e avaliação de possíveis atributos adicionais"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "WnNom3KLJB4N",
   "metadata": {
    "id": "WnNom3KLJB4N"
   },
   "source": [
    "* `fire_start_date`: data e hora de início do incêndio;\n",
    "* `discovered_date`: instante de descoberta do incêndio;\n",
    "* `reported_date`: data e hora que o incêndio foi reportado a primeira vez as autoridades;\n",
    "* `start_for_fire_date`: data e hora em que os primeiros recursos foram mobilizados para combate ao incêndio;\n",
    "* `fire_fighting_start_date`: data e hora de início de combate ao incêndio pelas autoridades;\n",
    "* `bh_fs_date`: A data e hora em que o Comandante do Incidente (IC) determinou que o incêndio florestal estava em estado de incêndio (BH);\n",
    "* `uc_fs_date`: A data e hora em que o Comandante do Incidente determinou que o incêndio florestal estava no estado de fogo sob controle (UC);\n",
    "* `ex_fs_date`: Esta é a data e hora em que o Comandante do Incidente determinou que o incêndio florestal estava completamente extinto (EX)."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "CX-ayE7PPfo-",
   "metadata": {
    "id": "CX-ayE7PPfo-"
   },
   "source": [
    "**Atributos possíveis**: \n",
    "1. Dia, mês e hora baseado em: `fire_start_date`;\n",
    "2. Turno baseado em: `fire_start_date`;\n",
    "3. Tempo de início de combate ao incêndio: `fire_fighting_start_date` - `fire_start_date`;\n",
    "4. Tempo de controle do incêndio: `uc_fs_date` - `bh_fs_date`;\n",
    "5. Tempo de extinção do incêndio: `ex_fs_date` - `bh_fs_date`;\n",
    "6. Tempo total do incêndio: `ex_fs_date` - `fire_start_date`;\n",
    "7. Área florestal na qual está localizado o local de ignição do fogo. Indicado pela primeira letra de `fire_number`:\n",
    "  * C–Calgary\n",
    "  * E–Edson\n",
    "  * H–High Level\n",
    "  * G–Grande Prairie\n",
    "  * L–Lac La Biche\n",
    "  * M–Fort McMurray\n",
    "  * P–Peace River\n",
    "  * R–Rocky\n",
    "  * S–Slave Lake\n",
    "  * W–Whitecourt"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3FkVvo6cxgmY",
   "metadata": {
    "id": "3FkVvo6cxgmY"
   },
   "source": [
    "### Ajuste dos Tipos de Variáveis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "Lr6HpAqEnctS",
   "metadata": {
    "id": "Lr6HpAqEnctS"
   },
   "outputs": [],
   "source": [
    "date_columns = ['fire_start_date', 'discovered_date', 'reported_date', 'start_for_fire_date', 'fire_fighting_start_date',\n",
    "                'bh_fs_date', 'uc_fs_date', 'ex_fs_date']\n",
    "\n",
    "for date in date_columns:\n",
    "  df_96_18[date] = pd.to_datetime(df_96_18[date])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "faefc04a",
   "metadata": {
    "id": "faefc04a"
   },
   "outputs": [],
   "source": [
    "df_96_18.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "SkZrTGGXv9BV",
   "metadata": {
    "id": "SkZrTGGXv9BV"
   },
   "source": [
    "### Remoção de Colunas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4lFqNzIGhzh6",
   "metadata": {
    "id": "4lFqNzIGhzh6"
   },
   "outputs": [],
   "source": [
    "removed_columns = ['fire_name', 'industry_identifier_desc', 'responsible_group_desc', 'activity_class', 'true_cause', \n",
    "                   'permit_detail_desc', 'other_fuel_type', 'to_fs_date', 'to_hectares', 'assessment_datetime', 'fire_year',\n",
    "                   'calendar_year', 'fire_fighting_start_size', 'discovered_date', 'fire_fighting_start_date',\n",
    "                   'reported_date', 'start_for_fire_date']\n",
    "df_96_18 = df_96_18.drop(removed_columns, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bWS5Tk8rz5J2",
   "metadata": {
    "id": "bWS5Tk8rz5J2"
   },
   "outputs": [],
   "source": [
    "df_96_18.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "VOFcNW96xFO1",
   "metadata": {
    "id": "VOFcNW96xFO1"
   },
   "source": [
    "### Remoção de Linhas (Registros)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "U5c7AZvv5_sJ",
   "metadata": {
    "id": "U5c7AZvv5_sJ"
   },
   "source": [
    "* Verificação do percentual de registros ausentes para cada atributo antes da aplicação do `dropna`. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "rrE1vS-b0apv",
   "metadata": {
    "id": "rrE1vS-b0apv"
   },
   "outputs": [],
   "source": [
    "(df_96_18.isna().sum()/df_96_18.shape[0])*100"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "O38-KGuA1E0g",
   "metadata": {
    "id": "O38-KGuA1E0g"
   },
   "outputs": [],
   "source": [
    "df_96_18.dropna(axis=0, how='any', inplace=True, subset=df_96_18.columns)\n",
    "df_96_18.reset_index(drop=True, inplace=True)\n",
    "df_96_18.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "suv7UzMo7f6p",
   "metadata": {
    "id": "suv7UzMo7f6p"
   },
   "source": [
    "### Nova verificação de Dados Ausentes após Remoção de Linhas e Colunas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7KDsrp-m7ZcE",
   "metadata": {
    "id": "7KDsrp-m7ZcE"
   },
   "outputs": [],
   "source": [
    "(df_96_18.isna().sum()/df_96_18.shape[0])*100"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "AQ-pG9OK6bTX",
   "metadata": {
    "id": "AQ-pG9OK6bTX"
   },
   "source": [
    "### Adição de novos atributos baseados nos atributos existentes no dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "MuonD1LG61tC",
   "metadata": {
    "id": "MuonD1LG61tC"
   },
   "source": [
    "* **Dia, Mês, Ano, Turno** e **Estações** extraídos do atributo `fire_start_date`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "oog3jfLcxYLe",
   "metadata": {
    "id": "oog3jfLcxYLe"
   },
   "outputs": [],
   "source": [
    "# Separando dia, mês e ano\n",
    "df_96_18['fire_start_day'] = df_96_18['fire_start_date'].apply(lambda dt: dt.day)      \n",
    "df_96_18['fire_start_month'] = df_96_18['fire_start_date'].apply(lambda dt: dt.month)      \n",
    "df_96_18['fire_start_year'] = df_96_18['fire_start_date'].apply(lambda dt: dt.year)      "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "B7xdB7covVJs",
   "metadata": {
    "id": "B7xdB7covVJs"
   },
   "outputs": [],
   "source": [
    "# Criando feature de turnos\n",
    "def define_day_period(dt):\n",
    "  hour = dt.strftime(\"%H:%M:%S\")\n",
    "  if (hour >= '00:00') and (hour <= '12:00'):\n",
    "    return 'morning'\n",
    "  elif (hour > '12:00') and (hour <= '18:00'):\n",
    "    return 'afternoon'\n",
    "  else:\n",
    "    return 'night'\n",
    "  \n",
    "df_96_18['day_period'] = df_96_18['fire_start_date'].apply(lambda dt: define_day_period(dt))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "46fiK3_F8mtn",
   "metadata": {
    "id": "46fiK3_F8mtn"
   },
   "source": [
    "**Estações do ano**: \n",
    "* Primavera – março a maio. `É quando a neve começa a derreter e fica mais quente.`<br>\n",
    "* Verão – Junho a Agosto/Setembro. `As temperaturas diurnas são geralmente superiores a 20 graus Celsius e podem atingir 30 graus Celsius ou mais em julho.`<br>\n",
    "* Outono – setembro a novembro.<br>\n",
    "* Inverno – dezembro a fevereiro ou mais tarde."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "wvexgqs4dCGX",
   "metadata": {
    "id": "wvexgqs4dCGX"
   },
   "outputs": [],
   "source": [
    "# Criando feature de estações do ano\n",
    "def define_season(dt):\n",
    "  month = dt.month\n",
    "  if (month >= 3) and (month <= 5):\n",
    "    return 'spring'\n",
    "  elif (month >= 6) and (month <= 8):\n",
    "    return 'summer'\n",
    "  elif (month >= 9) and (month <= 11):\n",
    "    return 'autumn'\n",
    "  else:\n",
    "    return 'winter'\n",
    "  \n",
    "df_96_18['seasons'] = df_96_18['fire_start_date'].apply(lambda dt: define_season(dt))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "19p16d3aT4jp",
   "metadata": {
    "id": "19p16d3aT4jp"
   },
   "outputs": [],
   "source": [
    "# Criando feature de tempo de controle \n",
    "df_96_18['control_time'] = (df_96_18['uc_fs_date'] - df_96_18['fire_start_date']).apply(lambda t: (t.components[0]*24*60) + (t.components[1]*60) + t.components[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "jX5CvjU9UXQv",
   "metadata": {
    "id": "jX5CvjU9UXQv"
   },
   "outputs": [],
   "source": [
    "# Criando feature de tempo de extinção\n",
    "df_96_18['extinction_time'] = (df_96_18['ex_fs_date'] - df_96_18['fire_start_date']).apply(lambda t: (t.components[0]*24*60) + (t.components[1]*60) + t.components[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "IRYxt-FQVwuQ",
   "metadata": {
    "id": "IRYxt-FQVwuQ"
   },
   "outputs": [],
   "source": [
    "# Criando feature delta de extinção: intervalo de tempo entre o controle do fogo e a extinção\n",
    "df_96_18['extinction_delta'] = df_96_18['extinction_time'] - df_96_18['control_time']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "gMzpC0Hq1WdY",
   "metadata": {
    "id": "gMzpC0Hq1WdY"
   },
   "outputs": [],
   "source": [
    "# Criando feature eficiência de extinção: tempo de extinçao do incêndio / (area total queimada * 10000)\n",
    "# A multiplicação por 10000 converte hectares em metros quadrados\n",
    "df_96_18['extinction_efficiency'] = round(df_96_18['extinction_time'] / (df_96_18['ex_hectares'] * 10000), 4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fPfJeW5myZ00",
   "metadata": {
    "id": "fPfJeW5myZ00"
   },
   "outputs": [],
   "source": [
    "# Criando feature causa do fogo: natural ou humana\n",
    "df_96_18['fire_cause'] = df_96_18['general_cause_desc'].apply(lambda c: 'natural' if c == 'Lightning' else 'human' )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3YRW-_9TWGS9",
   "metadata": {
    "id": "3YRW-_9TWGS9"
   },
   "outputs": [],
   "source": [
    "# Criando feature de área de proteção florestal\n",
    "def define_forest_protection_area(l):\n",
    "  cod = str(l[0:3]).lower().strip()\n",
    "  if (cod == 'cwf') or (cod[1:] == '01'):\n",
    "    return 'Calgary'\n",
    "  elif (cod == 'ewf') or (cod[1:] == '02'):\n",
    "    return 'Edson'\n",
    "  elif (cod == 'hwf') or (cod[1:] == '03'):\n",
    "    return 'High Level'\n",
    "  elif (cod == 'gwf') or (cod[1:] == '04'):\n",
    "    return 'Grande Prairie'\n",
    "  elif (cod == 'lwf') or (cod[1:] == '05'):\n",
    "    return 'Lac La Biche'\n",
    "  elif (cod == 'mwf'):\n",
    "    return 'Fort McMurray'\n",
    "  elif (cod == 'pwf') or (cod[1:] == '06'):\n",
    "    return 'Peace River'\n",
    "  elif (cod == 'rwf') or (cod[1:] == '07'):\n",
    "    return 'Rocky'\n",
    "  elif (cod == 'swf') or (cod[1:] == '08'):\n",
    "    return 'Slave Lake'\n",
    "  elif (cod == 'wwf') or (cod[1:] == '09'):\n",
    "    return 'Whitecourt' \n",
    "  else:\n",
    "    return 'Not Defined'\n",
    "  \n",
    "df_96_18['forest_protection_area'] = df_96_18['fire_number'].apply(lambda l: define_forest_protection_area(l))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "CVT8NpHEjBUL",
   "metadata": {
    "id": "CVT8NpHEjBUL"
   },
   "outputs": [],
   "source": [
    "df_96_18['forest_protection_area'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "FT3fD640aE3L",
   "metadata": {
    "id": "FT3fD640aE3L"
   },
   "outputs": [],
   "source": [
    "df_96_18.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "PO5FeOW_aa9a",
   "metadata": {
    "id": "PO5FeOW_aa9a"
   },
   "outputs": [],
   "source": [
    "# Novas features criadas\n",
    "df_96_18[['fire_start_day','fire_start_month','fire_start_year','day_period','seasons','control_time','extinction_time', 'ex_hectares', 'extinction_delta', 'extinction_efficiency', 'forest_protection_area']].sample(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "qwgOv4tvIG9H",
   "metadata": {
    "id": "qwgOv4tvIG9H"
   },
   "outputs": [],
   "source": [
    "df_96_18.describe().T"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "daZVLVzuFjXp",
   "metadata": {
    "id": "daZVLVzuFjXp"
   },
   "source": [
    "#### Corrigindo Inconsistências: tempo de controle zero e tempo de extinção zero"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "Q-Dq1vLnXfZ1",
   "metadata": {
    "id": "Q-Dq1vLnXfZ1"
   },
   "outputs": [],
   "source": [
    "df_96_18[df_96_18['extinction_time'] == 0][['fire_start_date', 'fire_number', 'extinction_time']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "OvfN98HSZVfv",
   "metadata": {
    "id": "OvfN98HSZVfv"
   },
   "outputs": [],
   "source": [
    "df_96_18[df_96_18['control_time'] == 0]['control_time'].count()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8vHIiBSdGRVF",
   "metadata": {
    "id": "8vHIiBSdGRVF"
   },
   "source": [
    "ATENÇÃO!!!\n",
    "* Os registros com tempo de extinção zero estão contidos nos intervalo de registros com tempo de controle zero. Portanto, basta excluir (drop) os registros referentes ao tempo de controle.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "RqkqTaek3veL",
   "metadata": {
    "id": "RqkqTaek3veL"
   },
   "outputs": [],
   "source": [
    "# Excluindo registros com tempo de controle igual a zero\n",
    "zero_reg_control_time_index = list(df_96_18[df_96_18['control_time'] == 0].index)\n",
    "\n",
    "df_96_18.drop(zero_reg_control_time_index, axis=0, inplace=True)\n",
    "\n",
    "df_96_18.reset_index(drop=True, inplace=True)\n",
    "df_96_18.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "AAx0lLmLF5og",
   "metadata": {
    "id": "AAx0lLmLF5og"
   },
   "source": [
    "### Análise das Variáveis Numéricas do Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "OKj39NnTN3Cz",
   "metadata": {
    "id": "OKj39NnTN3Cz"
   },
   "outputs": [],
   "source": [
    "df_96_18.describe().T"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "k8OVM3TSMb6d",
   "metadata": {
    "id": "k8OVM3TSMb6d"
   },
   "source": [
    "* Verificação do registro com eficiência de extinção zero"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "taeKz-sXIO9L",
   "metadata": {
    "id": "taeKz-sXIO9L"
   },
   "outputs": [],
   "source": [
    "df_96_18[df_96_18['extinction_efficiency'] == 0].T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "yuIGqMAKKhVi",
   "metadata": {
    "id": "yuIGqMAKKhVi"
   },
   "outputs": [],
   "source": [
    "df_96_18.loc[26004, 'extinction_time']/(df_96_18.loc[26004, 'ex_hectares']*10000)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6OqjU_cxzBZ5",
   "metadata": {
    "id": "6OqjU_cxzBZ5"
   },
   "source": [
    "### Análise das Variáveis Categóricas do Dataset "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7DD9xW07FeH-",
   "metadata": {
    "id": "7DD9xW07FeH-"
   },
   "outputs": [],
   "source": [
    "df_96_18.describe(include='object').T"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2RJPN7nnorC3",
   "metadata": {
    "id": "2RJPN7nnorC3"
   },
   "source": [
    "* Ranking das áreas de proteção ambiental mais afetadas pelos incêndios"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ownOWyWUo639",
   "metadata": {
    "id": "ownOWyWUo639"
   },
   "outputs": [],
   "source": [
    "dict_forest_protection_area_ranking = {'forest_protection_area':df_96_18['forest_protection_area'].value_counts().index, 'ranking':df_96_18['forest_protection_area'].value_counts().values}\n",
    "\n",
    "df_forest_protection_area_ranking = pd.DataFrame(dict_forest_protection_area_ranking)\n",
    "df_forest_protection_area_ranking"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ioR-HTGBl8Rs",
   "metadata": {
    "id": "ioR-HTGBl8Rs"
   },
   "source": [
    "* Ranking das vegetações mais atingidas pelos incêndios"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "Qymk6va9kB5K",
   "metadata": {
    "id": "Qymk6va9kB5K"
   },
   "outputs": [],
   "source": [
    "dict_vegetation_cover_ranking = {'vegetation_code':df_96_18['fuel_type'].value_counts().index, 'ranking':df_96_18['fuel_type'].value_counts().values}\n",
    "\n",
    "df_vegetation_cover_ranking = pd.DataFrame(dict_vegetation_cover_ranking)\n",
    "df_vegetation_cover_ranking"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bZGTtmd4mJh0",
   "metadata": {
    "id": "bZGTtmd4mJh0"
   },
   "source": [
    "* Ranking das estações do ano"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "Rn2LVcYlmxpy",
   "metadata": {
    "id": "Rn2LVcYlmxpy"
   },
   "outputs": [],
   "source": [
    "dict_season_ranking = {'season':df_96_18['seasons'].value_counts().index, 'ranking':df_96_18['seasons'].value_counts().values}\n",
    "\n",
    "df_season_ranking = pd.DataFrame(dict_season_ranking)\n",
    "df_season_ranking"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "-wSqqe8UGUk5",
   "metadata": {
    "id": "-wSqqe8UGUk5"
   },
   "source": [
    "### Boxplots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "od4EiL4RIWvZ",
   "metadata": {
    "id": "od4EiL4RIWvZ"
   },
   "outputs": [],
   "source": [
    "df_96_18.boxplot(column=['ex_hectares'], figsize=(10,8))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6OAd-81sGc8p",
   "metadata": {
    "id": "6OAd-81sGc8p"
   },
   "outputs": [],
   "source": [
    "df_96_18.boxplot(column=['extinction_delta'], figsize=(10,8))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9q927aCNHwyN",
   "metadata": {
    "id": "9q927aCNHwyN"
   },
   "outputs": [],
   "source": [
    "df_96_18.boxplot(column=['extinction_time'], figsize=(10,8))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "L_CfblwbFNgg",
   "metadata": {
    "id": "L_CfblwbFNgg"
   },
   "source": [
    "### Correlação"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "RpkqgJ-TFMxK",
   "metadata": {
    "id": "RpkqgJ-TFMxK"
   },
   "outputs": [],
   "source": [
    "round(df_96_18.corr(), 4)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "KD1lIHKxiz5h",
   "metadata": {
    "id": "KD1lIHKxiz5h"
   },
   "source": [
    "### Exportando dataset pré-processado"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4jXjYsyuz2-5",
   "metadata": {
    "id": "4jXjYsyuz2-5"
   },
   "outputs": [],
   "source": [
    "# df_96_18.to_csv('df_alberta_96_18.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8jy8mMEug_3_",
   "metadata": {
    "id": "8jy8mMEug_3_"
   },
   "source": [
    "### Preparação dos dados para os Modelos de Machine Learning"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "j86JX2dbHaI9",
   "metadata": {
    "id": "j86JX2dbHaI9"
   },
   "source": [
    "#### Divisão dos dados entre variáveis previsoras (features) e variável alvo (target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bdYy0xtVvSqf",
   "metadata": {
    "id": "bdYy0xtVvSqf"
   },
   "outputs": [],
   "source": [
    "df_96_18.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "NYJmGvRmQn1z",
   "metadata": {
    "id": "NYJmGvRmQn1z"
   },
   "outputs": [],
   "source": [
    "# Cria novo df para os modelos de ML \n",
    "df_ml = df_96_18.drop(['fire_number', 'general_cause_desc', 'det_agent', 'fire_start_date', 'bh_fs_date', 'uc_fs_date', 'ex_fs_date'], axis=1)\n",
    "df_ml.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "pp1U3XvFHnF5",
   "metadata": {
    "id": "pp1U3XvFHnF5"
   },
   "outputs": [],
   "source": [
    "# Divisão entre 'features' e 'target'\n",
    "\n",
    "X = df_ml.drop(['fire_cause'], axis=1)\n",
    "y = df_ml['fire_cause']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "jy6xupXj2kMZ",
   "metadata": {
    "id": "jy6xupXj2kMZ"
   },
   "outputs": [],
   "source": [
    "y.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "UaQeS_PVHos_",
   "metadata": {
    "id": "UaQeS_PVHos_"
   },
   "outputs": [],
   "source": [
    "# Armazenando os nomes das variáveis categóricas e numéricas em listas\n",
    "cat_cols = X.select_dtypes(include=\"object\").columns.to_list()\n",
    "num_cols = X.select_dtypes(exclude=\"object\").columns.to_list()\n",
    "\n",
    "print('Colunas Categóricas: ', cat_cols)\n",
    "print('Colunas Numéricas: ', num_cols)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4GxiWPFGHsJo",
   "metadata": {
    "id": "4GxiWPFGHsJo"
   },
   "outputs": [],
   "source": [
    "# Armazenando os nomes das features e das classes em listas\n",
    "df_target_names = list(y.unique())\n",
    "\n",
    "df_features_names = X.columns.to_list()\n",
    "\n",
    "print('Nº de Classes:', len(df_target_names))\n",
    "print('Classes:', df_target_names)\n",
    "print('\\nNº de Features:', len(df_features_names))\n",
    "print('Features:', df_features_names)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "JAFb78NBuw_q",
   "metadata": {
    "id": "JAFb78NBuw_q"
   },
   "source": [
    "## Implementação dos Modelos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b3dzTRxfJaYm",
   "metadata": {
    "id": "b3dzTRxfJaYm"
   },
   "outputs": [],
   "source": [
    "from sklearn import model_selection\n",
    "from sklearn.model_selection import RepeatedStratifiedKFold\n",
    "from sklearn.model_selection import RandomizedSearchCV\n",
    "from sklearn.model_selection import StratifiedKFold\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import cross_val_score, cross_validate\n",
    "from sklearn.metrics import classification_report, accuracy_score\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn import tree\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "oPN1ILOTtJeV",
   "metadata": {
    "id": "oPN1ILOTtJeV"
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import roc_curve\n",
    "from sklearn.preprocessing import LabelEncoder, MinMaxScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "KNuyndNxJ5nK",
   "metadata": {
    "id": "KNuyndNxJ5nK"
   },
   "outputs": [],
   "source": [
    "seed = 1275"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "PZtZxQLLdCPs",
   "metadata": {
    "id": "PZtZxQLLdCPs"
   },
   "source": [
    "### Funções para avaliação dos modelos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "E88Cc1FDc-x0",
   "metadata": {
    "id": "E88Cc1FDc-x0"
   },
   "outputs": [],
   "source": [
    "def evaluate(model, X, y, cv):\n",
    "    cv_results = cross_validate(\n",
    "        model,\n",
    "        X,\n",
    "        y,\n",
    "        cv=cv,\n",
    "        scoring=['accuracy', 'f1_micro', 'roc_auc'],\n",
    "    )\n",
    "    acc = cv_results[\"test_accuracy\"]\n",
    "    f1_m = cv_results[\"test_f1_micro\"]\n",
    "    auc = cv_results[\"test_roc_auc\"]\n",
    "    \n",
    "    print(\n",
    "        f\"Accuracy:     {acc.mean():.3f} +/- {acc.std():.3f}\\n\"\n",
    "        f\"F1_micro:     {f1_m.mean():.3f} +/- {f1_m.std():.3f}\\n\"\n",
    "        f\"Roc_auc:      {auc.mean():.3f} +/- {auc.std():.3f}\\n\"\n",
    "    )\n",
    "    #print(cv_results.keys())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "oIu17KSdfxWg",
   "metadata": {
    "id": "oIu17KSdfxWg"
   },
   "outputs": [],
   "source": [
    "def roc_curve_function(X_test, y_test, y_predicted, model, rn=False):\n",
    "  # calculate accuracy\n",
    "  acc = np.average(y_predicted == y_test)\n",
    "  print('Accuracy:', acc)\n",
    "\n",
    "  # calculate AUC\n",
    "  if rn == True: # Neural Network models\n",
    "    y_scores = np.reshape(model.predict(X_test), len(X_test))\n",
    "    auc = roc_auc_score(y_test,y_scores)\n",
    "    fpr, tpr, thresholds = roc_curve(y_test, y_scores)\n",
    "  else: # ML models\n",
    "    y_scores = model.predict_proba(X_test)\n",
    "    auc = roc_auc_score(y_test,y_scores[:,1])\n",
    "    fpr, tpr, thresholds = roc_curve(y_test, y_scores[:,1])\n",
    "  \n",
    "  print('AUC: ' + str(auc))\n",
    "\n",
    "  # plot ROC curve\n",
    "  fig = plt.figure(figsize=(6, 4))\n",
    "  # print(thresholds)\n",
    "\n",
    "  # Plot the diagonal 50% line\n",
    "  plt.plot([0, 1], [0, 1], 'k--')\n",
    "  plt.axis([0, 1, 0, 1])\n",
    "  # Plot the FPR and TPR achieved by our model\n",
    "  plt.plot(fpr, tpr)\n",
    "  plt.xlabel('False Positive Rate')\n",
    "  plt.ylabel('True Positive Rate')\n",
    "  plt.title('ROC Curve')\n",
    "\n",
    "  plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c1kBvaD4JBCw",
   "metadata": {
    "id": "c1kBvaD4JBCw"
   },
   "source": [
    "### Árvore de Decisão (Decision Tree)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "VIy5Hc3-N_AN",
   "metadata": {
    "id": "VIy5Hc3-N_AN"
   },
   "outputs": [],
   "source": [
    "le_tree = LabelEncoder()\n",
    "\n",
    "for atr in cat_cols:\n",
    "  X[atr] = le_tree.fit_transform(X[atr])\n",
    "\n",
    "X_tree = np.array(X)\n",
    "y_tree = le_tree.fit_transform(y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "pH7kkLXn3SJ7",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "pH7kkLXn3SJ7",
    "outputId": "a565dd35-988a-450f-ead1-9dbf42712255"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['human', 'natural'], dtype=object)"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "le_tree.classes_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "NVna91h63b1m",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "NVna91h63b1m",
    "outputId": "87775cdc-8f53-494a-d363-bcbf7fc6d0db"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['human', 'natural']"
      ]
     },
     "execution_count": 86,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "list(le_tree.inverse_transform([0, 1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "u7M76jSVJE33",
   "metadata": {
    "id": "u7M76jSVJE33"
   },
   "outputs": [],
   "source": [
    "# Divisao da base de dados em treinamento, validacao e teste\n",
    "X_train, X_test, y_train, y_test = train_test_split(X_tree, y_tree, test_size=0.25, stratify=y_tree, random_state=seed) # shuffle=False\n",
    "\n",
    "# Instanciando o modelo\n",
    "decisionTree = DecisionTreeClassifier(random_state=seed)\n",
    "\n",
    "# K-fold estratificado com k = 10\n",
    "cv = model_selection.StratifiedKFold(n_splits=10)\n",
    "\n",
    "# definição dos parâmetros para a árvore de decisão\n",
    "param_distributions = { 'criterion': ['gini', 'entropy'],\n",
    "                        'min_samples_split': [7,8,9,10,12,15,18,20,22,24],\n",
    "                        'max_depth': [2,3,4,5,6,7,8,9,10],\n",
    "                        'min_samples_leaf': [3,4,5,6,7,8,9,10,12,14,16,18]\n",
    "                      }\n",
    "\n",
    "# define random search for decision tree\n",
    "rnd_search_tree = RandomizedSearchCV( estimator=decisionTree, \n",
    "                                      param_distributions = param_distributions, \n",
    "                                      n_iter=25, scoring='accuracy', \n",
    "                                      n_jobs=-1, cv=cv, random_state=seed\n",
    "                                    )\n",
    "\n",
    "# execute search\n",
    "result_tree = rnd_search_tree.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7v1LCmT2JeJH",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "7v1LCmT2JeJH",
    "outputId": "e9f6a8e3-ce09-4d8f-9fcd-1ede38207510"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "=========Random Search Results for TREE==========\n",
      "Best Score: 0.9228268348747382\n",
      "Best Hyperparameters: {'min_samples_split': 8, 'min_samples_leaf': 10, 'max_depth': 9, 'criterion': 'entropy'}\n",
      "\n",
      "Desempenho médio da Árvore de Decisão:\n",
      "Arvore de Decisão: 0.922827 (0.005728)\n",
      "\n",
      "Acuracia Árvore de Decisão: Treinamento 0.9320618126131143  Teste 0.922466592427617\n",
      "\n",
      "Classification report:\n",
      "               precision    recall  f1-score   support\n",
      "\n",
      "       human       0.93      0.92      0.93      3918\n",
      "     natural       0.91      0.92      0.92      3266\n",
      "\n",
      "    accuracy                           0.92      7184\n",
      "   macro avg       0.92      0.92      0.92      7184\n",
      "weighted avg       0.92      0.92      0.92      7184\n",
      "\n",
      "Confusion matrix:\n",
      " [[3622  296]\n",
      " [ 261 3005]]\n"
     ]
    }
   ],
   "source": [
    "# summarize result for decision tree\n",
    "print('=========Random Search Results for TREE==========')\n",
    "print('Best Score: %s' % result_tree.best_score_)\n",
    "print('Best Hyperparameters: %s' % result_tree.best_params_)\n",
    "\n",
    "# Instanciando e avaliando o modelo\n",
    "decisionTree = DecisionTreeClassifier(**result_tree.best_params_, random_state=seed)\n",
    "\n",
    "model_dt = decisionTree.fit(X_train, y_train)\n",
    "y_predicted_tree = decisionTree.predict(X_test)\n",
    "\n",
    "print('\\nDesempenho médio da Árvore de Decisão:')\n",
    "\n",
    "cv_results = model_selection.cross_val_score(decisionTree, X_train, y_train, cv=cv, scoring='accuracy')\n",
    "# cv_models_results['decisionTree'] = cv_results\n",
    "\n",
    "name = 'Arvore de Decisão'\n",
    "msg = \"%s: %f (%f)\" % (name, cv_results.mean(), cv_results.std())\n",
    "print(msg)\n",
    "\n",
    "print(\"\\nAcuracia Árvore de Decisão: Treinamento\",  decisionTree.score(X_train, y_train),\" Teste\", decisionTree.score(X_test, y_test))\n",
    "print(\"\\nClassification report:\\n\", classification_report(le_tree.inverse_transform(y_test), le_tree.inverse_transform(y_predicted_tree)))\n",
    "print(\"Confusion matrix:\\n\", confusion_matrix(y_test, y_predicted_tree))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5782f14a",
   "metadata": {},
   "source": [
    "#### Representação Gráfica da Matrix de Confusão da Árvore de Decisão"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "mrpZJGqpS6Fb",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 279
    },
    "id": "mrpZJGqpS6Fb",
    "outputId": "1754e3bd-ff21-4dea-b53b-ca9d153f5c63"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "tree_confusion_matrix = pd.crosstab(le_tree.inverse_transform(y_test), le_tree.inverse_transform(y_predicted_tree), rownames=['Valor Atual'], colnames=['Valor Predito'])\n",
    "\n",
    "sn.heatmap(tree_confusion_matrix, annot=True, fmt=\"d\", cmap='inferno', linewidths=.5)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "09501ace",
   "metadata": {},
   "source": [
    "#### Funções de avaliação para a Árvore de Decisão"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "54lIKt0rdfwM",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "54lIKt0rdfwM",
    "outputId": "a6622043-543c-484e-edd2-04d4a5bb4725"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:     0.911 +/- 0.029\n",
      "F1_micro:     0.911 +/- 0.029\n",
      "Roc_auc:      0.956 +/- 0.031\n",
      "\n"
     ]
    }
   ],
   "source": [
    "evaluate(decisionTree, X_tree, y_tree, cv=cv)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "jvAggZuUslDj",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 332
    },
    "id": "jvAggZuUslDj",
    "outputId": "43fe7d38-3aed-4733-f0b1-0e57dc15d978"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.922466592427617\n",
      "AUC: 0.9720416736609372\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": [
    "roc_curve_function(X_test, y_test, y_predicted_tree, model_dt)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "de6eb35a",
   "metadata": {},
   "source": [
    "#### Features mais importantes segundo algoritmo da Árvore de Decisão"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fL4oFmBbJfo0",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "fL4oFmBbJfo0",
    "outputId": "8787e1a5-522a-4d2c-f248-13cecfc666da"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('fuel_type', 0.5557140828882152)\n",
      "('seasons', 0.24876216832469472)\n",
      "('fire_origin', 0.034683987118505434)\n",
      "('fire_start_month', 0.03402451742313796)\n",
      "('det_agent_type', 0.019852379732278914)\n",
      "('fire_location_latitude', 0.01847545797337875)\n",
      "('extinction_delta', 0.016655048683130502)\n",
      "('control_time', 0.015740845927350897)\n",
      "('weather_conditions_over_fire', 0.01325284764237778)\n",
      "('fire_start_day', 0.00967792968337146)\n",
      "('extinction_efficiency', 0.007353310395323477)\n",
      "('initial_action_by', 0.0072116361920390335)\n",
      "('fire_start_year', 0.00494145770991138)\n",
      "('day_period', 0.003449198173859898)\n",
      "('extinction_time', 0.002948430998361691)\n",
      "('fire_location_longitude', 0.002710592001891612)\n",
      "('fire_position_on_slope', 0.0022148218858424307)\n",
      "('forest_protection_area', 0.001107167718245088)\n",
      "('assessment_hectares', 0.0006019664859843819)\n",
      "('uc_hectares', 0.0004645330889897286)\n",
      "('fire_type', 0.00015761995310946858)\n",
      "('current_size', 0.0)\n",
      "('size_class', 0.0)\n",
      "('bh_hectares', 0.0)\n",
      "('ex_hectares', 0.0)\n"
     ]
    }
   ],
   "source": [
    "# Features mais importantes\n",
    "for feat in sorted(zip(df_features_names, decisionTree.feature_importances_), reverse=True, key=lambda x:x[1]):\n",
    "  print(feat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "IaDw3mcjga1Y",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 833
    },
    "id": "IaDw3mcjga1Y",
    "outputId": "a03236b1-5318-4a5c-e20e-bb759992f5e3"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-13f8889c-ee09-4655-9d85-590005ea4359\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <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>features</th>\n",
       "      <th>feature_importances_decision_tree</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>fuel_type</td>\n",
       "      <td>0.555714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>seasons</td>\n",
       "      <td>0.248762</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>fire_origin</td>\n",
       "      <td>0.034684</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>fire_start_month</td>\n",
       "      <td>0.034025</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>det_agent_type</td>\n",
       "      <td>0.019852</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>fire_location_latitude</td>\n",
       "      <td>0.018475</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>extinction_delta</td>\n",
       "      <td>0.016655</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>control_time</td>\n",
       "      <td>0.015741</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>weather_conditions_over_fire</td>\n",
       "      <td>0.013253</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>fire_start_day</td>\n",
       "      <td>0.009678</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>extinction_efficiency</td>\n",
       "      <td>0.007353</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>initial_action_by</td>\n",
       "      <td>0.007212</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>fire_start_year</td>\n",
       "      <td>0.004941</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>day_period</td>\n",
       "      <td>0.003449</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>extinction_time</td>\n",
       "      <td>0.002948</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>fire_location_longitude</td>\n",
       "      <td>0.002711</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>fire_position_on_slope</td>\n",
       "      <td>0.002215</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>forest_protection_area</td>\n",
       "      <td>0.001107</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>assessment_hectares</td>\n",
       "      <td>0.000602</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>uc_hectares</td>\n",
       "      <td>0.000465</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>fire_type</td>\n",
       "      <td>0.000158</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>ex_hectares</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>current_size</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>size_class</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>bh_hectares</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-13f8889c-ee09-4655-9d85-590005ea4359')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-13f8889c-ee09-4655-9d85-590005ea4359 button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-13f8889c-ee09-4655-9d85-590005ea4359');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "                        features  feature_importances_decision_tree\n",
       "0                      fuel_type                           0.555714\n",
       "1                        seasons                           0.248762\n",
       "2                    fire_origin                           0.034684\n",
       "3               fire_start_month                           0.034025\n",
       "4                 det_agent_type                           0.019852\n",
       "5         fire_location_latitude                           0.018475\n",
       "6               extinction_delta                           0.016655\n",
       "7                   control_time                           0.015741\n",
       "8   weather_conditions_over_fire                           0.013253\n",
       "9                 fire_start_day                           0.009678\n",
       "10         extinction_efficiency                           0.007353\n",
       "11             initial_action_by                           0.007212\n",
       "12               fire_start_year                           0.004941\n",
       "13                    day_period                           0.003449\n",
       "14               extinction_time                           0.002948\n",
       "15       fire_location_longitude                           0.002711\n",
       "16        fire_position_on_slope                           0.002215\n",
       "17        forest_protection_area                           0.001107\n",
       "18           assessment_hectares                           0.000602\n",
       "19                   uc_hectares                           0.000465\n",
       "20                     fire_type                           0.000158\n",
       "21                   ex_hectares                           0.000000\n",
       "22                  current_size                           0.000000\n",
       "23                    size_class                           0.000000\n",
       "24                   bh_hectares                           0.000000"
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dict_feat_importance_decision_tree = {'features':df_features_names, 'feature_importances_decision_tree':decisionTree.feature_importances_}\n",
    "\n",
    "df_feat_importance_decision_tree = pd.DataFrame(dict_feat_importance_decision_tree)\n",
    "df_feat_importance_decision_tree.sort_values(by=['feature_importances_decision_tree'], ascending=False, inplace=True)\n",
    "df_feat_importance_decision_tree.reset_index(inplace=True, drop=True)\n",
    "df_feat_importance_decision_tree"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "jIA2BStzgaeH",
   "metadata": {
    "id": "jIA2BStzgaeH"
   },
   "outputs": [],
   "source": [
    "# df_feat_importance_decision_tree.to_csv('df_feat_importances_decision_tree.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a3bkNKUPupTW",
   "metadata": {
    "id": "a3bkNKUPupTW"
   },
   "source": [
    "### Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "63760287",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-23T01:18:19.304126Z",
     "start_time": "2022-04-23T01:18:17.388990Z"
    },
    "id": "63760287"
   },
   "outputs": [],
   "source": [
    "# Instanciando o modelo\n",
    "rfc = RandomForestClassifier(random_state=seed)\n",
    "\n",
    "# definição dos parâmetros para a árvore de decisão\n",
    "param_distributions = { \n",
    "                        'n_estimators': [10,11,12,15,17,18,19,20,22,24,26,28,30],\n",
    "                        'criterion': ['gini', 'entropy'],\n",
    "                        'max_depth': [10,12,14,15,16,17,18,19,20,21,22],\n",
    "                        'min_samples_split': [7,8,9,10,12,13,14,15,16,17,18],\n",
    "                        'min_samples_leaf': [6,7,8,9,10,12,15,18],\n",
    "                        'max_features':['auto', 'sqrt', 'log2']\n",
    "                      }\n",
    "\n",
    "# define random search for decision tree\n",
    "rnd_search_rfc = RandomizedSearchCV(  estimator=rfc, \n",
    "                                      param_distributions = param_distributions, \n",
    "                                      n_iter=25, scoring='accuracy', \n",
    "                                      n_jobs=-1, cv=cv, random_state=seed\n",
    "                                    )\n",
    "# execute search\n",
    "result_rfc = rnd_search_rfc.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a6544088",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-23T01:18:19.904961Z",
     "start_time": "2022-04-23T01:18:19.305928Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "a6544088",
    "outputId": "14c62dcf-d538-4720-c837-cb4dffdf1b78"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "=========Random Search Results for RandomForest==========\n",
      "Best Score: 0.9360062862596324\n",
      "Best Hyperparameters: {'n_estimators': 26, 'min_samples_split': 10, 'min_samples_leaf': 7, 'max_features': 'sqrt', 'max_depth': 19, 'criterion': 'entropy'}\n",
      "\n",
      "Desempenho médio do RandomForest:\n",
      "RandomForest: 0.936006 (0.002512)\n",
      "\n",
      "Acuracia do Random Forest: Treinamento 0.9595340851083577  Teste 0.9416759465478842\n",
      "\n",
      "Classification report:\n",
      "               precision    recall  f1-score   support\n",
      "\n",
      "       human       0.95      0.94      0.95      3918\n",
      "     natural       0.93      0.94      0.94      3266\n",
      "\n",
      "    accuracy                           0.94      7184\n",
      "   macro avg       0.94      0.94      0.94      7184\n",
      "weighted avg       0.94      0.94      0.94      7184\n",
      "\n",
      "Confusion matrix:\n",
      " [[3699  219]\n",
      " [ 200 3066]]\n"
     ]
    }
   ],
   "source": [
    "# summarize result for random forest\n",
    "print('=========Random Search Results for RandomForest==========')\n",
    "print('Best Score: %s' % result_rfc.best_score_)\n",
    "print('Best Hyperparameters: %s' % result_rfc.best_params_)\n",
    "\n",
    "# Instanciando e avaliando o modelo\n",
    "RFC = RandomForestClassifier(**result_rfc.best_params_, random_state=seed)\n",
    "\n",
    "model_rfc = RFC.fit(X_train, y_train)\n",
    "y_predicted_rfc = RFC.predict(X_test)\n",
    "\n",
    "print('\\nDesempenho médio do RandomForest:')\n",
    "\n",
    "cv_results = model_selection.cross_val_score(RFC, X_train, y_train, cv=cv, scoring='accuracy')\n",
    "\n",
    "name = 'RandomForest'\n",
    "msg = \"%s: %f (%f)\" % (name, cv_results.mean(), cv_results.std())\n",
    "print(msg)\n",
    "\n",
    "print(\"\\nAcuracia do Random Forest: Treinamento\",  RFC.score(X_train, y_train),\" Teste\", RFC.score(X_test, y_test))\n",
    "print(\"\\nClassification report:\\n\", classification_report(le_tree.inverse_transform(y_test), le_tree.inverse_transform(y_predicted_rfc)))\n",
    "print(\"Confusion matrix:\\n\", confusion_matrix(y_test, y_predicted_rfc))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3038276d",
   "metadata": {},
   "source": [
    "#### Representação Gráfica da Matrix de Confusão do Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3718b8ab",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-23T01:18:20.063917Z",
     "start_time": "2022-04-23T01:18:19.906934Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 279
    },
    "id": "3718b8ab",
    "outputId": "a12c3438-f4bc-4ff9-97ae-fbe812009926"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "rnd_frt_confusion_matrix = pd.crosstab(le_tree.inverse_transform(y_test), le_tree.inverse_transform(y_predicted_rfc), rownames=['Valor Atual'], colnames=['Valor Predito'])\n",
    "\n",
    "sn.heatmap(rnd_frt_confusion_matrix, annot=True, fmt=\"d\", cmap='inferno', linewidths=.5)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "148dc315",
   "metadata": {},
   "source": [
    "#### Funções de avaliação para o Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "rVKKffVgd3wJ",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "rVKKffVgd3wJ",
    "outputId": "5d9e14d1-222e-4afb-adad-8a303393181e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:     0.925 +/- 0.025\n",
      "F1_micro:     0.925 +/- 0.025\n",
      "Roc_auc:      0.978 +/- 0.008\n",
      "\n"
     ]
    }
   ],
   "source": [
    "evaluate(RFC, X_tree, y_tree, cv=cv)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "JzZHjzCHvxFa",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 332
    },
    "id": "JzZHjzCHvxFa",
    "outputId": "725614b8-eabb-4237-b3b5-79557161e37a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.9416759465478842\n",
      "AUC: 0.9825755138952319\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": [
    "roc_curve_function(X_test, y_test, y_predicted_rfc, model_rfc)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e942a366",
   "metadata": {},
   "source": [
    "#### Features mais importantes segundo algoritmo Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b7bff7f0",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-23T01:18:20.079901Z",
     "start_time": "2022-04-23T01:18:20.065910Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "b7bff7f0",
    "outputId": "c2e70aad-a196-44ee-fc28-000fd497113c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('fuel_type', 0.27074371384205864)\n",
      "('seasons', 0.16590935831054032)\n",
      "('fire_origin', 0.14038258874451165)\n",
      "('fire_start_month', 0.0769671742874027)\n",
      "('fire_location_latitude', 0.05339784795816045)\n",
      "('initial_action_by', 0.045184720700076474)\n",
      "('det_agent_type', 0.035796931201564464)\n",
      "('weather_conditions_over_fire', 0.03539982696769712)\n",
      "('control_time', 0.027267815399884515)\n",
      "('extinction_delta', 0.021491104461561016)\n",
      "('fire_location_longitude', 0.016299708764551175)\n",
      "('extinction_efficiency', 0.016020850177991024)\n",
      "('fire_start_day', 0.015866052937867827)\n",
      "('fire_start_year', 0.014201663394408943)\n",
      "('extinction_time', 0.01342709187722432)\n",
      "('forest_protection_area', 0.011024800952788022)\n",
      "('day_period', 0.0068300568966496306)\n",
      "('ex_hectares', 0.00665900829535797)\n",
      "('uc_hectares', 0.006270106133215212)\n",
      "('current_size', 0.00518927265313853)\n",
      "('bh_hectares', 0.005121954006757641)\n",
      "('assessment_hectares', 0.004346636608257374)\n",
      "('fire_type', 0.0030803363627532744)\n",
      "('fire_position_on_slope', 0.0024177386489744744)\n",
      "('size_class', 0.0007036404166071905)\n"
     ]
    }
   ],
   "source": [
    "# Features mais importantes\n",
    "for feat in sorted(zip(df_features_names, RFC.feature_importances_), reverse=True, key=lambda x:x[1]):\n",
    "  print(feat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0dzKpHRiam9I",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 833
    },
    "id": "0dzKpHRiam9I",
    "outputId": "83ba694c-4ee5-470a-868c-14dbddfbb910"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-588a8d28-037d-4f6a-80e3-853bb002274c\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <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>features</th>\n",
       "      <th>feature_importances_RFC</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>fuel_type</td>\n",
       "      <td>0.270744</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>seasons</td>\n",
       "      <td>0.165909</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>fire_origin</td>\n",
       "      <td>0.140383</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>fire_start_month</td>\n",
       "      <td>0.076967</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>fire_location_latitude</td>\n",
       "      <td>0.053398</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>initial_action_by</td>\n",
       "      <td>0.045185</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>det_agent_type</td>\n",
       "      <td>0.035797</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>weather_conditions_over_fire</td>\n",
       "      <td>0.035400</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>control_time</td>\n",
       "      <td>0.027268</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>extinction_delta</td>\n",
       "      <td>0.021491</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>fire_location_longitude</td>\n",
       "      <td>0.016300</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>extinction_efficiency</td>\n",
       "      <td>0.016021</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>fire_start_day</td>\n",
       "      <td>0.015866</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>fire_start_year</td>\n",
       "      <td>0.014202</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>extinction_time</td>\n",
       "      <td>0.013427</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>forest_protection_area</td>\n",
       "      <td>0.011025</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>day_period</td>\n",
       "      <td>0.006830</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>ex_hectares</td>\n",
       "      <td>0.006659</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>uc_hectares</td>\n",
       "      <td>0.006270</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>current_size</td>\n",
       "      <td>0.005189</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>bh_hectares</td>\n",
       "      <td>0.005122</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>assessment_hectares</td>\n",
       "      <td>0.004347</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>fire_type</td>\n",
       "      <td>0.003080</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>fire_position_on_slope</td>\n",
       "      <td>0.002418</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>size_class</td>\n",
       "      <td>0.000704</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-588a8d28-037d-4f6a-80e3-853bb002274c')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-588a8d28-037d-4f6a-80e3-853bb002274c button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-588a8d28-037d-4f6a-80e3-853bb002274c');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "                        features  feature_importances_RFC\n",
       "0                      fuel_type                 0.270744\n",
       "1                        seasons                 0.165909\n",
       "2                    fire_origin                 0.140383\n",
       "3               fire_start_month                 0.076967\n",
       "4         fire_location_latitude                 0.053398\n",
       "5              initial_action_by                 0.045185\n",
       "6                 det_agent_type                 0.035797\n",
       "7   weather_conditions_over_fire                 0.035400\n",
       "8                   control_time                 0.027268\n",
       "9               extinction_delta                 0.021491\n",
       "10       fire_location_longitude                 0.016300\n",
       "11         extinction_efficiency                 0.016021\n",
       "12                fire_start_day                 0.015866\n",
       "13               fire_start_year                 0.014202\n",
       "14               extinction_time                 0.013427\n",
       "15        forest_protection_area                 0.011025\n",
       "16                    day_period                 0.006830\n",
       "17                   ex_hectares                 0.006659\n",
       "18                   uc_hectares                 0.006270\n",
       "19                  current_size                 0.005189\n",
       "20                   bh_hectares                 0.005122\n",
       "21           assessment_hectares                 0.004347\n",
       "22                     fire_type                 0.003080\n",
       "23        fire_position_on_slope                 0.002418\n",
       "24                    size_class                 0.000704"
      ]
     },
     "execution_count": 100,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dict_feat_importance_RFC = {'features':df_features_names, 'feature_importances_RFC':RFC.feature_importances_}\n",
    "\n",
    "df_feat_importance_RFC = pd.DataFrame(dict_feat_importance_RFC)\n",
    "df_feat_importance_RFC.sort_values(by=['feature_importances_RFC'], ascending=False, inplace=True)\n",
    "df_feat_importance_RFC.reset_index(inplace=True, drop=True)\n",
    "df_feat_importance_RFC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "jDy5XgraZcxx",
   "metadata": {
    "id": "jDy5XgraZcxx"
   },
   "outputs": [],
   "source": [
    "# df_feat_importance_RFC.to_csv('df_feat_importances_RFC.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7RNbnxZ4VbfN",
   "metadata": {
    "id": "7RNbnxZ4VbfN"
   },
   "source": [
    "### XGBoost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "D5UmdFgnnzDG",
   "metadata": {
    "id": "D5UmdFgnnzDG"
   },
   "outputs": [],
   "source": [
    "#!pip install xgboost==1.2.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "wAIguLKuVd-u",
   "metadata": {
    "id": "wAIguLKuVd-u"
   },
   "outputs": [],
   "source": [
    "import xgboost\n",
    "from xgboost import XGBClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "G9kB6NvmmZf1",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 36
    },
    "id": "G9kB6NvmmZf1",
    "outputId": "7214919e-e9ef-47c4-f3de-c0adcf54cfb9"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.google.colaboratory.intrinsic+json": {
       "type": "string"
      },
      "text/plain": [
       "'0.90'"
      ]
     },
     "execution_count": 104,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xgboost.__version__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fRqgSmjJ1xnQ",
   "metadata": {
    "id": "fRqgSmjJ1xnQ"
   },
   "outputs": [],
   "source": [
    "# Divisao da base de dados em treinamento, validacao e teste\n",
    "X_train_xgb, X_test_xgb, y_train_xgb, y_test_xgb = train_test_split(X_tree, y_tree, test_size=0.25, stratify=y_tree, random_state=seed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "DCUm3WpyVfB4",
   "metadata": {
    "id": "DCUm3WpyVfB4"
   },
   "outputs": [],
   "source": [
    "xgb = XGBClassifier(\n",
    "                    booster='gbtree',\n",
    "                    objective='binary:logistic', max_depth=15,\n",
    "                    learning_rate=0.1, n_estimators=100,\n",
    "                    random_state=seed, n_jobs=-1, tree_method='gpu_hist', gpu_id=0\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "NdDod3E06R1p",
   "metadata": {
    "id": "NdDod3E06R1p"
   },
   "source": [
    "* Obs1.: n_estimators=100: Popular among ensemble methods, `n_estimators` is the number of boosted trees in the model. Increasing this number while decreasing learning_rate can lead to more robust results."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "gZTlWAR6b06M",
   "metadata": {
    "id": "gZTlWAR6b06M"
   },
   "source": [
    "#### Avaliação do Modelo antes da busca de hiperparâmetros - XGBoost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "eyOADTFS7Tku",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "eyOADTFS7Tku",
    "outputId": "c89401c6-6652-4b99-867b-262780c72be2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:     0.897 +/- 0.059\n",
      "F1_micro:     0.897 +/- 0.059\n",
      "Roc_auc:      0.964 +/- 0.030\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Avaliação do modelo antes da busca de parâmetros\n",
    "evaluate(xgb, X_tree, y_tree, cv=cv)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7rdH6QLKXb9r",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "7rdH6QLKXb9r",
    "outputId": "bbd5081d-17a5-421b-d029-6972afc9b7b1"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Classification report:\n",
      "               precision    recall  f1-score   support\n",
      "\n",
      "       human       0.96      0.95      0.95      3918\n",
      "     natural       0.94      0.95      0.94      3266\n",
      "\n",
      "    accuracy                           0.95      7184\n",
      "   macro avg       0.95      0.95      0.95      7184\n",
      "weighted avg       0.95      0.95      0.95      7184\n",
      "\n",
      "Confusion matrix:\n",
      " [[3706  212]\n",
      " [ 171 3095]]\n"
     ]
    }
   ],
   "source": [
    "model = xgb.fit(X_train_xgb, y_train_xgb)\n",
    "y_predicted = xgb.predict(X_test_xgb)\n",
    "\n",
    "print(\"\\nClassification report:\\n\", classification_report(le_tree.inverse_transform(y_test_xgb), le_tree.inverse_transform(y_predicted)))\n",
    "print(\"Confusion matrix:\\n\", confusion_matrix(y_test_xgb, y_predicted))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "twearpoQcHTN",
   "metadata": {
    "id": "twearpoQcHTN"
   },
   "source": [
    "#### Busca dos melhores hiperparâmetros - XGBoost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "isjcrTXk16MJ",
   "metadata": {
    "id": "isjcrTXk16MJ"
   },
   "outputs": [],
   "source": [
    "PARAM_DISTRIBUTIONS = [\n",
    "    {\n",
    "      'n_estimators': [75, 85, 100, 125, 150, 175, 200, 225],\n",
    "      'learning_rate':[0.01, 0.1, 0.2, 0.3],\n",
    "      'max_depth':[5, 6, 7, 8, 9, 10, 12, 14],\n",
    "      'booster':['gbtree', 'gblinear'],\n",
    "      'objective':['binary:logistic','binary:logitraw'],\n",
    "      'gamma':[0, 0.1, 0.5, 1.0, 1.5],\n",
    "      'min_child_weight':[1, 2, 3, 4, 5],\n",
    "      'subsample':[0.5, 0.7, 0.8, 0.9, 1],\n",
    "      'colsample_bytree':[0.5, 0.7, 0.8, 0.9, 1],\n",
    "      'colsample_bynode':[0.5, 0.7, 0.8, 0.9, 1],\n",
    "      'colsample_bylevel':[0.5, 0.7, 0.8, 0.9, 1]\n",
    "    }\n",
    "  ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "QpbOMpY5181Y",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "QpbOMpY5181Y",
    "outputId": "72337279-f64e-4840-e8f8-ece1bad3c389"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomizedSearchCV(cv=StratifiedKFold(n_splits=10, random_state=None, shuffle=False),\n",
       "                   estimator=XGBClassifier(gpu_id=0, max_depth=15, n_jobs=-1,\n",
       "                                           random_state=1275,\n",
       "                                           tree_method='gpu_hist'),\n",
       "                   n_iter=25, n_jobs=-1,\n",
       "                   param_distributions=[{'booster': ['gbtree', 'gblinear'],\n",
       "                                         'colsample_bylevel': [0.5, 0.7, 0.8,\n",
       "                                                               0.9, 1],\n",
       "                                         'colsample_bynode': [0.5, 0.7, 0.8,\n",
       "                                                              0.9, 1],\n",
       "                                         'colsample_bytree': [0.5, 0.7, 0.8,\n",
       "                                                              0.9, 1],\n",
       "                                         'gamma': [0, 0.1, 0.5, 1.0, 1.5],\n",
       "                                         'learning_rate': [0.01, 0.1, 0.2, 0.3],\n",
       "                                         'max_depth': [5, 6, 7, 8, 9, 10, 12,\n",
       "                                                       14],\n",
       "                                         'min_child_weight': [1, 2, 3, 4, 5],\n",
       "                                         'n_estimators': [75, 85, 100, 125, 150,\n",
       "                                                          175, 200, 225],\n",
       "                                         'objective': ['binary:logistic',\n",
       "                                                       'binary:logitraw'],\n",
       "                                         'subsample': [0.5, 0.7, 0.8, 0.9, 1]}],\n",
       "                   random_state=1275, scoring='accuracy')"
      ]
     },
     "execution_count": 141,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rnd_search = RandomizedSearchCV(estimator=xgb, \n",
    "                                param_distributions = PARAM_DISTRIBUTIONS, \n",
    "                                n_iter=25, scoring= 'accuracy',\n",
    "                                n_jobs=-1, cv=cv, random_state=seed\n",
    "                              )\n",
    "rnd_search.fit(X_train_xgb, y_train_xgb)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2fZhyJnpcUd-",
   "metadata": {
    "id": "2fZhyJnpcUd-"
   },
   "source": [
    "#### Avaliação do Modelo com os melhores hiperparâmetros - XGBoost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "VgIx46dJsZVK",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "VgIx46dJsZVK",
    "outputId": "77bab2e1-9e7c-4492-962d-75b212a3b9b2"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'booster': 'gbtree',\n",
       " 'colsample_bylevel': 0.7,\n",
       " 'colsample_bynode': 0.7,\n",
       " 'colsample_bytree': 0.7,\n",
       " 'gamma': 1.5,\n",
       " 'learning_rate': 0.1,\n",
       " 'max_depth': 8,\n",
       " 'min_child_weight': 3,\n",
       " 'n_estimators': 150,\n",
       " 'objective': 'binary:logistic',\n",
       " 'subsample': 0.8}"
      ]
     },
     "execution_count": 142,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rnd_search.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "uc9SVSOh3PC9",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "uc9SVSOh3PC9",
    "outputId": "a43f7b2c-682a-4a04-acd7-500416c0932f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<bound method XGBModel.get_params of XGBClassifier(colsample_bylevel=0.7, colsample_bynode=0.7, colsample_bytree=0.7,\n",
      "              gamma=1.5, max_depth=8, min_child_weight=3, n_estimators=150,\n",
      "              random_state=1275, subsample=0.8)>\n",
      "\n",
      "\n",
      "Accuracy:     0.865 +/- 0.092\n",
      "F1_micro:     0.865 +/- 0.092\n",
      "Roc_auc:      0.948 +/- 0.053\n",
      "\n"
     ]
    }
   ],
   "source": [
    "hyper_xgb = XGBClassifier(**rnd_search.best_params_, random_state=seed)\n",
    "print(hyper_xgb.get_params)\n",
    "print('\\n')\n",
    "\n",
    "evaluate(hyper_xgb, X_tree, y_tree, cv=cv)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "jIqvg44SVhtS",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "jIqvg44SVhtS",
    "outputId": "ac1fe8c8-6ccc-42fa-dabb-a3f2fd1359f4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Classification report:\n",
      "               precision    recall  f1-score   support\n",
      "\n",
      "       human       0.96      0.95      0.95      3918\n",
      "     natural       0.94      0.95      0.95      3266\n",
      "\n",
      "    accuracy                           0.95      7184\n",
      "   macro avg       0.95      0.95      0.95      7184\n",
      "weighted avg       0.95      0.95      0.95      7184\n",
      "\n",
      "Confusion matrix:\n",
      " [[3727  191]\n",
      " [ 169 3097]]\n"
     ]
    }
   ],
   "source": [
    "model_hyper = hyper_xgb.fit(X_train_xgb, y_train_xgb)\n",
    "y_predicted_h = hyper_xgb.predict(X_test_xgb)\n",
    "\n",
    "print(\"\\nClassification report:\\n\", classification_report(le_tree.inverse_transform(y_test_xgb), le_tree.inverse_transform(y_predicted_h)))\n",
    "print(\"Confusion matrix:\\n\", confusion_matrix(y_test_xgb, y_predicted_h))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "61meL-vUwqyN",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 332
    },
    "id": "61meL-vUwqyN",
    "outputId": "627bcc4b-f8a0-4d0d-b554-de2c7323fce0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.9498886414253898\n",
      "AUC: 0.9885018100703116\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": [
    "roc_curve_function(X_test_xgb, y_test_xgb, y_predicted_h, model_hyper)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "SdfEzWljNZzb",
   "metadata": {
    "id": "SdfEzWljNZzb"
   },
   "source": [
    "## Redes Neurais - MLP"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "HtFxEcg-a5_r",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "HtFxEcg-a5_r",
    "outputId": "91bb0281-5cde-4511-a48a-f9bfbc9091c4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
      "Collecting tensorflow-addons\n",
      "  Downloading tensorflow_addons-0.17.1-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.1 MB)\n",
      "\u001b[K     |████████████████████████████████| 1.1 MB 31.6 MB/s \n",
      "\u001b[?25hRequirement already satisfied: packaging in /usr/local/lib/python3.7/dist-packages (from tensorflow-addons) (21.3)\n",
      "Requirement already satisfied: typeguard>=2.7 in /usr/local/lib/python3.7/dist-packages (from tensorflow-addons) (2.7.1)\n",
      "Requirement already satisfied: pyparsing!=3.0.5,>=2.0.2 in /usr/local/lib/python3.7/dist-packages (from packaging->tensorflow-addons) (3.0.9)\n",
      "Installing collected packages: tensorflow-addons\n",
      "Successfully installed tensorflow-addons-0.17.1\n"
     ]
    }
   ],
   "source": [
    "#!pip install tensorflow-addons"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "XUiiymuEdkQJ",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-19T00:01:16.279708Z",
     "start_time": "2022-04-19T00:01:10.577867Z"
    },
    "id": "XUiiymuEdkQJ"
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "import tensorflow_addons as tfa"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "R7LgWrcgUqYP",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-19T00:01:17.450340Z",
     "start_time": "2022-04-19T00:01:16.281702Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "R7LgWrcgUqYP",
    "outputId": "014b366a-d03b-4c62-9520-15c8b19fe75c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensorflow Version:  2.8.2\n",
      "GPUs Available:  1\n",
      "Tensorflow Adons Version:  0.17.1\n"
     ]
    }
   ],
   "source": [
    "print(\"Tensorflow Version: \", tf.__version__)\n",
    "print(\"GPUs Available: \", len(tf.config.list_physical_devices('GPU')))\n",
    "print(\"Tensorflow Adons Version: \", tfa.__version__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "LGGdvgexbXXZ",
   "metadata": {
    "id": "LGGdvgexbXXZ"
   },
   "outputs": [],
   "source": [
    "# Aplicando o processamento de dados\n",
    "le_mlp = LabelEncoder()\n",
    "\n",
    "y_processed = le_mlp.fit_transform(y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bpLccivO7Rgl",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "bpLccivO7Rgl",
    "outputId": "ae58b83f-5c6d-440d-8141-f00ec621c72f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(28733, 98)"
      ]
     },
     "execution_count": 91,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_gdm = pd.get_dummies(X, columns=cat_cols, drop_first=True)\n",
    "X_gdm.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "lAFW9EsN9NqX",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 317
    },
    "id": "lAFW9EsN9NqX",
    "outputId": "c5ae313e-1f1b-4803-a8ed-7b74549569ca"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-f9b5c5af-001d-432a-b694-87d764b5b26a\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <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>assessment_hectares</th>\n",
       "      <th>current_size</th>\n",
       "      <th>fire_location_latitude</th>\n",
       "      <th>fire_location_longitude</th>\n",
       "      <th>bh_hectares</th>\n",
       "      <th>uc_hectares</th>\n",
       "      <th>ex_hectares</th>\n",
       "      <th>fire_start_day</th>\n",
       "      <th>fire_start_month</th>\n",
       "      <th>fire_start_year</th>\n",
       "      <th>...</th>\n",
       "      <th>seasons_winter</th>\n",
       "      <th>forest_protection_area_Edson</th>\n",
       "      <th>forest_protection_area_Fort McMurray</th>\n",
       "      <th>forest_protection_area_Grande Prairie</th>\n",
       "      <th>forest_protection_area_High Level</th>\n",
       "      <th>forest_protection_area_Lac La Biche</th>\n",
       "      <th>forest_protection_area_Peace River</th>\n",
       "      <th>forest_protection_area_Rocky</th>\n",
       "      <th>forest_protection_area_Slave Lake</th>\n",
       "      <th>forest_protection_area_Whitecourt</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.00</td>\n",
       "      <td>1.0</td>\n",
       "      <td>56.6914</td>\n",
       "      <td>-111.3831</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>1996</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.40</td>\n",
       "      <td>0.4</td>\n",
       "      <td>57.2789</td>\n",
       "      <td>-111.5762</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.4</td>\n",
       "      <td>19</td>\n",
       "      <td>5</td>\n",
       "      <td>1996</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.00</td>\n",
       "      <td>4.0</td>\n",
       "      <td>58.2523</td>\n",
       "      <td>-110.5540</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>24</td>\n",
       "      <td>5</td>\n",
       "      <td>1996</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.02</td>\n",
       "      <td>1.2</td>\n",
       "      <td>57.7477</td>\n",
       "      <td>-112.6786</td>\n",
       "      <td>1.2</td>\n",
       "      <td>1.2</td>\n",
       "      <td>1.2</td>\n",
       "      <td>29</td>\n",
       "      <td>5</td>\n",
       "      <td>1996</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.04</td>\n",
       "      <td>1.6</td>\n",
       "      <td>57.5501</td>\n",
       "      <td>-111.2103</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.8</td>\n",
       "      <td>1.6</td>\n",
       "      <td>30</td>\n",
       "      <td>5</td>\n",
       "      <td>1996</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 98 columns</p>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-f9b5c5af-001d-432a-b694-87d764b5b26a')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-f9b5c5af-001d-432a-b694-87d764b5b26a button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-f9b5c5af-001d-432a-b694-87d764b5b26a');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "   assessment_hectares  current_size  fire_location_latitude  \\\n",
       "0                 1.00           1.0                 56.6914   \n",
       "1                 0.40           0.4                 57.2789   \n",
       "2                 1.00           4.0                 58.2523   \n",
       "3                 0.02           1.2                 57.7477   \n",
       "4                 0.04           1.6                 57.5501   \n",
       "\n",
       "   fire_location_longitude  bh_hectares  uc_hectares  ex_hectares  \\\n",
       "0                -111.3831          1.0          1.0          1.0   \n",
       "1                -111.5762          0.4          0.4          0.4   \n",
       "2                -110.5540          4.0          4.0          4.0   \n",
       "3                -112.6786          1.2          1.2          1.2   \n",
       "4                -111.2103          0.8          0.8          1.6   \n",
       "\n",
       "   fire_start_day  fire_start_month  fire_start_year  ...  seasons_winter  \\\n",
       "0               4                 5             1996  ...               0   \n",
       "1              19                 5             1996  ...               0   \n",
       "2              24                 5             1996  ...               0   \n",
       "3              29                 5             1996  ...               0   \n",
       "4              30                 5             1996  ...               0   \n",
       "\n",
       "   forest_protection_area_Edson  forest_protection_area_Fort McMurray  \\\n",
       "0                             0                                     0   \n",
       "1                             0                                     0   \n",
       "2                             0                                     0   \n",
       "3                             0                                     0   \n",
       "4                             0                                     0   \n",
       "\n",
       "   forest_protection_area_Grande Prairie  forest_protection_area_High Level  \\\n",
       "0                                      0                                  0   \n",
       "1                                      0                                  0   \n",
       "2                                      0                                  0   \n",
       "3                                      0                                  0   \n",
       "4                                      0                                  0   \n",
       "\n",
       "   forest_protection_area_Lac La Biche  forest_protection_area_Peace River  \\\n",
       "0                                    0                                   0   \n",
       "1                                    0                                   0   \n",
       "2                                    0                                   0   \n",
       "3                                    0                                   0   \n",
       "4                                    0                                   0   \n",
       "\n",
       "   forest_protection_area_Rocky  forest_protection_area_Slave Lake  \\\n",
       "0                             0                                  0   \n",
       "1                             0                                  0   \n",
       "2                             0                                  0   \n",
       "3                             0                                  0   \n",
       "4                             0                                  0   \n",
       "\n",
       "   forest_protection_area_Whitecourt  \n",
       "0                                  0  \n",
       "1                                  0  \n",
       "2                                  0  \n",
       "3                                  0  \n",
       "4                                  0  \n",
       "\n",
       "[5 rows x 98 columns]"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_gdm.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "i3uT8N5_bwlQ",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "i3uT8N5_bwlQ",
    "outputId": "2fd3b190-a39e-455e-e3e1-2b0d50dfc18e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape do Conjunto de Treino:  (14366, 98)\n",
      "Shape das labels de Treino:  (14366,)\n",
      "Shape do Conjunto de Validação:  (7183, 98)\n",
      "Shape das labels de Validação:  (7183,)\n",
      "Shape do Conjunto de Teste:  (7184, 98)\n",
      "Shape das labels de Teste:  (7184,)\n"
     ]
    }
   ],
   "source": [
    "X_train_mlp, X_test_mlp, y_train_mlp, y_test_mlp = train_test_split(X_gdm, y_processed, test_size=0.25, stratify=y_processed, random_state=seed)\n",
    "X_train_mlp, X_valid_mlp, y_train_mlp, y_valid_mlp = train_test_split(X_train_mlp, y_train_mlp, test_size = 0.3333, stratify=y_train_mlp, random_state=seed)\n",
    "\n",
    "print('Shape do Conjunto de Treino: ', X_train_mlp.shape)\n",
    "print('Shape das labels de Treino: ', y_train_mlp.shape)\n",
    "print('Shape do Conjunto de Validação: ', X_valid_mlp.shape)\n",
    "print('Shape das labels de Validação: ', y_valid_mlp.shape)\n",
    "print('Shape do Conjunto de Teste: ', X_test_mlp.shape)\n",
    "print('Shape das labels de Teste: ', y_test_mlp.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "y0bbaMt8w3rZ",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "y0bbaMt8w3rZ",
    "outputId": "0804198e-3d0a-49d1-ca54-9fb7870b94b7"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "y train: [7835 6531]\n",
      "y valid: [3917 3266]\n",
      "y test: [3918 3266]\n"
     ]
    }
   ],
   "source": [
    "print(\"y train:\", np.bincount(y_train_mlp))\n",
    "print(\"y valid:\", np.bincount(y_valid_mlp))\n",
    "print(\"y test:\", np.bincount(y_test_mlp))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "FaDkMr9oUqYQ",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-19T00:01:17.465384Z",
     "start_time": "2022-04-19T00:01:17.451332Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "FaDkMr9oUqYQ",
    "outputId": "978e9428-c440-4e9e-b689-84ad363e8c98"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MinMaxScaler(feature_range=(-1, 1))"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Scaler = MinMaxScaler(feature_range=(-1,1))\n",
    "Scaler.fit(X_train_mlp[num_cols])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "MAoLWKnnAsdd",
   "metadata": {
    "id": "MAoLWKnnAsdd"
   },
   "outputs": [],
   "source": [
    "X_train_mlp.loc[:,num_cols] = Scaler.transform(X_train_mlp[num_cols])\n",
    "X_valid_mlp.loc[:,num_cols] = Scaler.transform(X_valid_mlp[num_cols])\n",
    "X_test_mlp.loc[:,num_cols] = Scaler.transform(X_test_mlp[num_cols])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2aJZBwlqUqYQ",
   "metadata": {
    "id": "2aJZBwlqUqYQ"
   },
   "source": [
    "* Definindo o modelo (MLP)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "lGCAMt-nNkJZ",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-19T00:01:19.368101Z",
     "start_time": "2022-04-19T00:01:17.481332Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "lGCAMt-nNkJZ",
    "outputId": "6770a152-8baa-4206-edc1-28c618128501"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " dense (Dense)               (None, 100)               9900      \n",
      "                                                                 \n",
      " dropout (Dropout)           (None, 100)               0         \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 50)                5050      \n",
      "                                                                 \n",
      " dropout_1 (Dropout)         (None, 50)                0         \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 10)                510       \n",
      "                                                                 \n",
      " dropout_2 (Dropout)         (None, 10)                0         \n",
      "                                                                 \n",
      " dense_3 (Dense)             (None, 1)                 11        \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 15,471\n",
      "Trainable params: 15,471\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "INPUT_SHAPE = (X_train_mlp.shape[1],)\n",
    "DROPOUT = 0.3\n",
    "\n",
    "model_mlp = keras.models.Sequential([\n",
    "                                      keras.layers.Dense(100, activation='relu', input_shape=INPUT_SHAPE),\n",
    "                                      keras.layers.Dropout(DROPOUT),\n",
    "                                      keras.layers.Dense(50, activation='relu'),\n",
    "                                      keras.layers.Dropout(DROPOUT),\n",
    "                                      keras.layers.Dense(10, activation='relu'),\n",
    "                                      keras.layers.Dropout(DROPOUT),\n",
    "                                      keras.layers.Dense(1, activation = \"sigmoid\")\n",
    "])\n",
    "\n",
    "model_mlp.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "O6_hPEvfUqYR",
   "metadata": {
    "id": "O6_hPEvfUqYR"
   },
   "source": [
    "* Definindo função de perda, otimizador e métricas;\n",
    "* Compilando o modelo."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "vLCXwWzaUqYR",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-19T00:01:19.398108Z",
     "start_time": "2022-04-19T00:01:19.369101Z"
    },
    "id": "vLCXwWzaUqYR"
   },
   "outputs": [],
   "source": [
    "from tensorflow_addons.metrics.f_scores import F1Score\n",
    "\n",
    "f1_score = F1Score(num_classes=2, average=\"micro\", threshold=0.5)\n",
    "\n",
    "model_mlp.compile(loss='binary_crossentropy', optimizer=\"Adam\", metrics= ['accuracy', keras.metrics.AUC(name='auc'), f1_score])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "smwGepduUqYR",
   "metadata": {
    "id": "smwGepduUqYR"
   },
   "source": [
    "* Definindo tamanho dos lotes (*batch size*) e o número de épocas."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "jfsNRmidgsq9",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-19T00:01:19.414164Z",
     "start_time": "2022-04-19T00:01:19.399100Z"
    },
    "id": "jfsNRmidgsq9"
   },
   "outputs": [],
   "source": [
    "BATCH_SIZE = 50    # Tamanho do batch de treinamento\n",
    "N_EPOCHS = 100"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "W2-w5h3KUqYR",
   "metadata": {
    "id": "W2-w5h3KUqYR"
   },
   "source": [
    "* Treinando o Modelo."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "uKH9lfFKgxTW",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-19T00:02:13.517138Z",
     "start_time": "2022-04-19T00:01:19.416130Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "uKH9lfFKgxTW",
    "outputId": "cd6862d8-0e05-4201-e6dd-a01e0f57dcb8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "288/288 [==============================] - 8s 12ms/step - loss: 0.3281 - accuracy: 0.8666 - auc: 0.9374 - f1_score: 0.8550 - val_loss: 0.1951 - val_accuracy: 0.9257 - val_auc: 0.9761 - val_f1_score: 0.9185\n",
      "Epoch 2/100\n",
      "288/288 [==============================] - 3s 9ms/step - loss: 0.2245 - accuracy: 0.9210 - auc: 0.9681 - f1_score: 0.9133 - val_loss: 0.1873 - val_accuracy: 0.9284 - val_auc: 0.9782 - val_f1_score: 0.9207\n",
      "Epoch 3/100\n",
      "288/288 [==============================] - 3s 12ms/step - loss: 0.2124 - accuracy: 0.9245 - auc: 0.9718 - f1_score: 0.9171 - val_loss: 0.1815 - val_accuracy: 0.9293 - val_auc: 0.9790 - val_f1_score: 0.9221\n",
      "Epoch 4/100\n",
      "288/288 [==============================] - 4s 13ms/step - loss: 0.2071 - accuracy: 0.9286 - auc: 0.9731 - f1_score: 0.9218 - val_loss: 0.1911 - val_accuracy: 0.9272 - val_auc: 0.9792 - val_f1_score: 0.9174\n",
      "Epoch 5/100\n",
      "288/288 [==============================] - 3s 11ms/step - loss: 0.1981 - accuracy: 0.9303 - auc: 0.9754 - f1_score: 0.9235 - val_loss: 0.1771 - val_accuracy: 0.9326 - val_auc: 0.9800 - val_f1_score: 0.9262\n",
      "Epoch 6/100\n",
      "288/288 [==============================] - 3s 10ms/step - loss: 0.1941 - accuracy: 0.9332 - auc: 0.9760 - f1_score: 0.9269 - val_loss: 0.1771 - val_accuracy: 0.9316 - val_auc: 0.9803 - val_f1_score: 0.9260\n",
      "Epoch 7/100\n",
      "288/288 [==============================] - 3s 9ms/step - loss: 0.1904 - accuracy: 0.9347 - auc: 0.9772 - f1_score: 0.9286 - val_loss: 0.1805 - val_accuracy: 0.9323 - val_auc: 0.9798 - val_f1_score: 0.9271\n",
      "Epoch 8/100\n",
      "288/288 [==============================] - 3s 9ms/step - loss: 0.1864 - accuracy: 0.9358 - auc: 0.9778 - f1_score: 0.9299 - val_loss: 0.1776 - val_accuracy: 0.9325 - val_auc: 0.9799 - val_f1_score: 0.9258\n",
      "Epoch 9/100\n",
      "288/288 [==============================] - 3s 9ms/step - loss: 0.1840 - accuracy: 0.9357 - auc: 0.9786 - f1_score: 0.9297 - val_loss: 0.1743 - val_accuracy: 0.9337 - val_auc: 0.9807 - val_f1_score: 0.9272\n",
      "Epoch 10/100\n",
      "288/288 [==============================] - 2s 8ms/step - loss: 0.1802 - accuracy: 0.9380 - auc: 0.9795 - f1_score: 0.9323 - val_loss: 0.1790 - val_accuracy: 0.9340 - val_auc: 0.9810 - val_f1_score: 0.9290\n",
      "Epoch 11/100\n",
      "288/288 [==============================] - 3s 9ms/step - loss: 0.1759 - accuracy: 0.9411 - auc: 0.9801 - f1_score: 0.9358 - val_loss: 0.1749 - val_accuracy: 0.9326 - val_auc: 0.9807 - val_f1_score: 0.9250\n",
      "Epoch 12/100\n",
      "288/288 [==============================] - 3s 9ms/step - loss: 0.1727 - accuracy: 0.9386 - auc: 0.9811 - f1_score: 0.9330 - val_loss: 0.1724 - val_accuracy: 0.9344 - val_auc: 0.9811 - val_f1_score: 0.9288\n",
      "Epoch 13/100\n",
      "288/288 [==============================] - 2s 5ms/step - loss: 0.1692 - accuracy: 0.9415 - auc: 0.9819 - f1_score: 0.9362 - val_loss: 0.1708 - val_accuracy: 0.9362 - val_auc: 0.9813 - val_f1_score: 0.9304\n",
      "Epoch 14/100\n",
      "288/288 [==============================] - 2s 5ms/step - loss: 0.1668 - accuracy: 0.9411 - auc: 0.9820 - f1_score: 0.9357 - val_loss: 0.1721 - val_accuracy: 0.9367 - val_auc: 0.9814 - val_f1_score: 0.9311\n",
      "Epoch 15/100\n",
      "288/288 [==============================] - 2s 5ms/step - loss: 0.1626 - accuracy: 0.9429 - auc: 0.9832 - f1_score: 0.9376 - val_loss: 0.1768 - val_accuracy: 0.9350 - val_auc: 0.9807 - val_f1_score: 0.9281\n",
      "Epoch 16/100\n",
      "288/288 [==============================] - 2s 5ms/step - loss: 0.1606 - accuracy: 0.9423 - auc: 0.9834 - f1_score: 0.9370 - val_loss: 0.1758 - val_accuracy: 0.9347 - val_auc: 0.9809 - val_f1_score: 0.9281\n",
      "Epoch 17/100\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1591 - accuracy: 0.9456 - auc: 0.9837 - f1_score: 0.9405 - val_loss: 0.1782 - val_accuracy: 0.9357 - val_auc: 0.9806 - val_f1_score: 0.9302\n",
      "Epoch 18/100\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1536 - accuracy: 0.9448 - auc: 0.9848 - f1_score: 0.9398 - val_loss: 0.1770 - val_accuracy: 0.9350 - val_auc: 0.9807 - val_f1_score: 0.9286\n",
      "Epoch 19/100\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1542 - accuracy: 0.9461 - auc: 0.9847 - f1_score: 0.9410 - val_loss: 0.1783 - val_accuracy: 0.9360 - val_auc: 0.9806 - val_f1_score: 0.9303\n",
      "Epoch 20/100\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1475 - accuracy: 0.9481 - auc: 0.9859 - f1_score: 0.9434 - val_loss: 0.1783 - val_accuracy: 0.9347 - val_auc: 0.9806 - val_f1_score: 0.9282\n",
      "\n",
      "Quantidade de épocas executadas:  20\n"
     ]
    }
   ],
   "source": [
    "callback_mlp = tf.keras.callbacks.EarlyStopping(monitor='val_loss', patience=7) \n",
    "# This callback will stop the training when there is no improvement in the loss for five consecutive epochs.\n",
    "\n",
    "history_mlp = model_mlp.fit(\n",
    "                            x=X_train_mlp, \n",
    "                            y=y_train_mlp, \n",
    "                            epochs=N_EPOCHS, \n",
    "                            batch_size=BATCH_SIZE, \n",
    "                            verbose='auto', \n",
    "                            validation_data=(X_valid_mlp,y_valid_mlp),\n",
    "                            callbacks=[callback_mlp],\n",
    "                          )\n",
    "print('\\nQuantidade de épocas executadas: ', len(history_mlp.history['loss']))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "shSJOta8UqYS",
   "metadata": {
    "id": "shSJOta8UqYS"
   },
   "source": [
    "* Plotando o histórico das métricas `accuracy` e `f1-score`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "dHnVHkTDjuFY",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-19T00:02:15.191726Z",
     "start_time": "2022-04-19T00:02:13.518139Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 651
    },
    "id": "dHnVHkTDjuFY",
    "outputId": "b33a6cc2-2009-4d9e-eec3-4cd2eebbaa4d"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plot_history(history):\n",
    "  hist = pd.DataFrame(history.history)\n",
    "  hist['epoch'] = history.epoch\n",
    "\n",
    "  plt.figure(figsize=(8,5))\n",
    "  plt.xlabel('Epoch')\n",
    "  plt.ylabel('accuracy')\n",
    "  plt.plot(hist['epoch'], hist['accuracy'],\n",
    "           label='Train accuracy')\n",
    "  plt.plot(hist['epoch'], hist['val_accuracy'],\n",
    "           label = 'Val accuracy')\n",
    "  # plt.ylim([0,5])\n",
    "  plt.legend()\n",
    "\n",
    "  plt.figure(figsize=(8,5))\n",
    "  plt.xlabel('Epoch')\n",
    "  plt.ylabel('F1 Score')\n",
    "  plt.plot(hist['epoch'], hist['f1_score'],\n",
    "           label='Train F1 Score')\n",
    "  plt.plot(hist['epoch'], hist['val_f1_score'],\n",
    "           label = 'Val F1 Score')\n",
    "  # plt.ylim([0,20])\n",
    "  plt.legend()\n",
    "  plt.show()\n",
    "\n",
    "plot_history(history_mlp)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6sezdwziNFtC",
   "metadata": {
    "id": "6sezdwziNFtC"
   },
   "source": [
    "### Predição e Métricas de Avaliação - MLP"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "TIcUO8-QUqYS",
   "metadata": {
    "id": "TIcUO8-QUqYS"
   },
   "source": [
    "* Avaliando o modelo treinado e exibindo as métricas obtidas a partir do conjunto de testes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "agCiF1QYl8ed",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-19T00:02:15.456903Z",
     "start_time": "2022-04-19T00:02:15.193716Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "agCiF1QYl8ed",
    "outputId": "445aba4f-5396-43c7-fbe6-69f92b656954"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "F1 Score: 0.933\n",
      "Accuracy: 0.939\n",
      "AUC: 0.981\n"
     ]
    }
   ],
   "source": [
    "loss, acc, auc, f1 = model_mlp.evaluate(X_test_mlp, y_test_mlp, batch_size=BATCH_SIZE, verbose='auto')\n",
    "\n",
    "print(\"F1 Score: {:5.3f}\".format(f1))\n",
    "print(\"Accuracy: {:5.3f}\".format(acc))\n",
    "print(\"AUC: {:5.3f}\".format(auc))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "MNq2URaNa07X",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-30T21:37:02.150291Z",
     "start_time": "2022-03-30T21:37:01.898444Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "MNq2URaNa07X",
    "outputId": "79cb4ac8-c593-4813-ae87-1b328b650ae2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1 1 0 ... 1 1 1]\n"
     ]
    }
   ],
   "source": [
    "y_predicted_mlp = np.reshape((model_mlp.predict(X_test_mlp) > 0.5).astype('int8'), len(X_test_mlp))\n",
    "print(y_predicted_mlp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ZMGh7kMHbM5e",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-30T21:37:06.421913Z",
     "start_time": "2022-03-30T21:37:06.401899Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "ZMGh7kMHbM5e",
    "outputId": "421149b0-fac3-4db6-dc2e-4c3d7e792280"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Classification report:\n",
      "               precision    recall  f1-score   support\n",
      "\n",
      "       human       0.94      0.95      0.94      3918\n",
      "     natural       0.94      0.93      0.93      3266\n",
      "\n",
      "    accuracy                           0.94      7184\n",
      "   macro avg       0.94      0.94      0.94      7184\n",
      "weighted avg       0.94      0.94      0.94      7184\n",
      "\n",
      "Confusion matrix:\n",
      " [[3712  206]\n",
      " [ 231 3035]]\n"
     ]
    }
   ],
   "source": [
    "print(\"\\nClassification report:\\n\", classification_report(le_mlp.inverse_transform(y_test_mlp), le_mlp.inverse_transform(y_predicted_mlp)))\n",
    "print(\"Confusion matrix:\\n\", confusion_matrix(y_test_mlp, y_predicted_mlp))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "132b855b",
   "metadata": {},
   "source": [
    "#### Representação Gráfica da Matrix de Confusão da Rede Neural MLP"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cW-kHvmtd329",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-30T21:38:12.681864Z",
     "start_time": "2022-03-30T21:38:12.546865Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 279
    },
    "id": "cW-kHvmtd329",
    "outputId": "3d8d68e5-b45c-4889-a5ea-a0f05b2feb9f"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Matriz de confusão com heatmap\n",
    "mlp_confusion_matrix = pd.crosstab(np.reshape(le_mlp.inverse_transform(y_test_mlp), len(y_test_mlp)), np.reshape(le_mlp.inverse_transform(y_predicted_mlp), len(y_predicted_mlp)), rownames=['Valor Atual'], colnames=['Valor Predito'])\n",
    "\n",
    "sn.heatmap(mlp_confusion_matrix, annot=True, fmt=\"d\", cmap='inferno', linewidths=.5)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b2138f19",
   "metadata": {},
   "source": [
    "#### Curva ROC e AUC para a Rede Neural MLP"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "kWVYXxi_tRAp",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 332
    },
    "id": "kWVYXxi_tRAp",
    "outputId": "e05c8c55-689c-41eb-fabb-edfcb36d1eb3"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.9391703786191536\n",
      "AUC: 0.9817193214104075\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": [
    "roc_curve_function(X_test_mlp, y_test_mlp, y_predicted_mlp, model_mlp, rn=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "YxL14-iHX83N",
   "metadata": {
    "id": "YxL14-iHX83N"
   },
   "source": [
    "### Ajuste dos Hiperparâmetros com keras tuner - Rede Neural MLP"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "X5cAo58ABL2k",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-25T15:15:47.927692Z",
     "start_time": "2022-03-25T15:15:45.841649Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "X5cAo58ABL2k",
    "outputId": "a9df8a4f-903f-4703-e0f5-4cc4820a1e14"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[?25l\r",
      "\u001b[K     |██▌                             | 10 kB 35.7 MB/s eta 0:00:01\r",
      "\u001b[K     |█████                           | 20 kB 41.2 MB/s eta 0:00:01\r",
      "\u001b[K     |███████▍                        | 30 kB 44.8 MB/s eta 0:00:01\r",
      "\u001b[K     |█████████▉                      | 40 kB 28.2 MB/s eta 0:00:01\r",
      "\u001b[K     |████████████▎                   | 51 kB 21.8 MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████▊                 | 61 kB 24.6 MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████████▏              | 71 kB 25.3 MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████████▋            | 81 kB 26.5 MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████████████          | 92 kB 28.6 MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████████████▌       | 102 kB 30.2 MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████████████████     | 112 kB 30.2 MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████████████████████▍  | 122 kB 30.2 MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████████████████████▉| 133 kB 30.2 MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████████████████████| 133 kB 30.2 MB/s \n",
      "\u001b[?25h"
     ]
    }
   ],
   "source": [
    "#!pip install -q -U keras-tuner"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "U1bmzz7KAzTz",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-31T11:51:52.218433Z",
     "start_time": "2022-03-31T11:51:52.118953Z"
    },
    "id": "U1bmzz7KAzTz"
   },
   "outputs": [],
   "source": [
    "import keras_tuner as kt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "z42BC7s1BbdV",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-31T11:51:54.989586Z",
     "start_time": "2022-03-31T11:51:54.975549Z"
    },
    "id": "z42BC7s1BbdV"
   },
   "outputs": [],
   "source": [
    "from tensorflow_addons.metrics.f_scores import F1Score\n",
    "METRICS_F1 = F1Score(num_classes=2, average=\"micro\", threshold=0.5)\n",
    "\n",
    "def model_mlp_builder(hp):\n",
    "  # Hiperparameters\n",
    "  INPUT_SHAPE = (X_train_mlp.shape[1],)\n",
    "\n",
    "  DROPOUT = 0.3\n",
    "  HP_HIDDEN_LAYER_1_UNITS = hp.Int('units_1', min_value=50, max_value=150, step=10)\n",
    "  HP_HIDDEN_LAYER_2_UNITS = hp.Int('units_2', min_value=30, max_value=80, step=10)\n",
    "  HP_HIDDEN_LAYER_3_UNITS = hp.Int('units_3', min_value=5, max_value=30, step=5)\n",
    "\n",
    "  model_mlp_tn = keras.models.Sequential([\n",
    "                                          keras.layers.Dense(units=HP_HIDDEN_LAYER_1_UNITS, activation='relu', input_shape=INPUT_SHAPE),\n",
    "                                          keras.layers.Dropout(DROPOUT),\n",
    "                                          keras.layers.Dense(units=HP_HIDDEN_LAYER_2_UNITS, activation='relu'),\n",
    "                                          keras.layers.Dropout(DROPOUT),\n",
    "                                          keras.layers.Dense(units=HP_HIDDEN_LAYER_3_UNITS, activation='relu'),\n",
    "                                          keras.layers.Dropout(DROPOUT),\n",
    "                                          keras.layers.Dense(1, activation = \"sigmoid\")\n",
    "  ])\n",
    "    \n",
    "  HP_LEARNING_RATE = hp.Choice('learning_rate', values=[1e-2, 1e-3, 1e-4])\n",
    "\n",
    "  model_mlp_tn.compile(\n",
    "      loss=\"binary_crossentropy\", \n",
    "      optimizer=keras.optimizers.Adam(learning_rate=HP_LEARNING_RATE), \n",
    "      metrics=['accuracy', keras.metrics.AUC(name='auc'), METRICS_F1]\n",
    "  )\n",
    "\n",
    "  return model_mlp_tn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "uJG7pu8_Fs0W",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-31T11:52:26.199052Z",
     "start_time": "2022-03-31T11:52:26.127859Z"
    },
    "id": "uJG7pu8_Fs0W"
   },
   "outputs": [],
   "source": [
    "MAX_EPOCHS = 100\n",
    "\n",
    "tuner_mlp = kt.Hyperband(\n",
    "                          model_mlp_builder,\n",
    "                          objective= kt.Objective('val_loss', direction='min'),\n",
    "                          max_epochs=MAX_EPOCHS,\n",
    "                          factor=3,\n",
    "                          directory='mlp_tuner_dir_1',\n",
    "                          project_name='mlp_tuner_val_loss'\n",
    "                        )\n",
    "#tuner_mlp.search_space_summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "EE2kaEBOGvnH",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-31T12:51:30.360740Z",
     "start_time": "2022-03-31T12:10:27.298949Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "EE2kaEBOGvnH",
    "outputId": "903ef083-a872-407a-9a71-bbc2449a90d6"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trial 254 Complete [00h 02m 39s]\n",
      "val_loss: 0.173521026968956\n",
      "\n",
      "Best val_loss So Far: 0.1702076643705368\n",
      "Total elapsed time: 01h 08m 29s\n",
      "INFO:tensorflow:Oracle triggered exit\n"
     ]
    }
   ],
   "source": [
    "stop_early = tf.keras.callbacks.EarlyStopping(monitor='val_loss', patience=6)\n",
    "\n",
    "BATCH_SIZE = 50    # Tamanho do batch de treinamento\n",
    "N_EPOCHS = 100\n",
    "\n",
    "tuner_mlp.search(\n",
    "                  x=X_train_mlp, \n",
    "                  y=y_train_mlp, \n",
    "                  epochs=N_EPOCHS, \n",
    "                  batch_size=BATCH_SIZE, \n",
    "                  #verbose='auto', \n",
    "                  validation_data=(X_valid_mlp,y_valid_mlp),\n",
    "                  callbacks=[stop_early],\n",
    "                )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "hcC6otY2O9Jb",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-31T12:52:00.702683Z",
     "start_time": "2022-03-31T12:52:00.696670Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "hcC6otY2O9Jb",
    "outputId": "b98245da-1738-44e8-fc5d-496ae48bc294"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "units_1 130\n",
      "units_2 60\n",
      "units_3 20\n",
      "learning_rate 0.001\n"
     ]
    }
   ],
   "source": [
    "# Get the optimal hyperparameters\n",
    "best_hps=tuner_mlp.get_best_hyperparameters(num_trials=1)[0]\n",
    "\n",
    "for h_param in ['units_1', 'units_2', 'units_3'] + ['learning_rate']:\n",
    "  print(h_param, best_hps.get(h_param))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "SHbpaeT7H4cm",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T19:34:40.274108Z",
     "start_time": "2022-04-01T19:31:00.072074Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "SHbpaeT7H4cm",
    "outputId": "5b470969-e61a-4114-d807-3553ca0ea591"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n",
      "288/288 [==============================] - 3s 8ms/step - loss: 0.2949 - accuracy: 0.8789 - auc: 0.9480 - f1_score: 0.8872 - val_loss: 0.1916 - val_accuracy: 0.9262 - val_auc: 0.9772 - val_f1_score: 0.9199\n",
      "Epoch 2/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.2113 - accuracy: 0.9242 - auc: 0.9713 - f1_score: 0.9171 - val_loss: 0.1856 - val_accuracy: 0.9265 - val_auc: 0.9787 - val_f1_score: 0.9190\n",
      "Epoch 3/50\n",
      "288/288 [==============================] - 2s 7ms/step - loss: 0.2016 - accuracy: 0.9272 - auc: 0.9738 - f1_score: 0.9204 - val_loss: 0.1802 - val_accuracy: 0.9294 - val_auc: 0.9796 - val_f1_score: 0.9231\n",
      "Epoch 4/50\n",
      "288/288 [==============================] - 2s 7ms/step - loss: 0.1890 - accuracy: 0.9321 - auc: 0.9771 - f1_score: 0.9258 - val_loss: 0.1773 - val_accuracy: 0.9312 - val_auc: 0.9801 - val_f1_score: 0.9245\n",
      "Epoch 5/50\n",
      "288/288 [==============================] - 3s 9ms/step - loss: 0.1850 - accuracy: 0.9333 - auc: 0.9779 - f1_score: 0.9273 - val_loss: 0.1813 - val_accuracy: 0.9280 - val_auc: 0.9801 - val_f1_score: 0.9232\n",
      "Epoch 6/50\n",
      "288/288 [==============================] - 3s 10ms/step - loss: 0.1784 - accuracy: 0.9361 - auc: 0.9794 - f1_score: 0.9304 - val_loss: 0.1725 - val_accuracy: 0.9344 - val_auc: 0.9810 - val_f1_score: 0.9288\n",
      "Epoch 7/50\n",
      "288/288 [==============================] - 3s 12ms/step - loss: 0.1738 - accuracy: 0.9382 - auc: 0.9802 - f1_score: 0.9327 - val_loss: 0.1751 - val_accuracy: 0.9330 - val_auc: 0.9805 - val_f1_score: 0.9265\n",
      "Epoch 8/50\n",
      "288/288 [==============================] - 3s 11ms/step - loss: 0.1708 - accuracy: 0.9384 - auc: 0.9810 - f1_score: 0.9328 - val_loss: 0.1721 - val_accuracy: 0.9358 - val_auc: 0.9812 - val_f1_score: 0.9302\n",
      "Epoch 9/50\n",
      "288/288 [==============================] - 3s 12ms/step - loss: 0.1650 - accuracy: 0.9415 - auc: 0.9820 - f1_score: 0.9364 - val_loss: 0.1794 - val_accuracy: 0.9336 - val_auc: 0.9807 - val_f1_score: 0.9261\n",
      "Epoch 10/50\n",
      "288/288 [==============================] - 3s 10ms/step - loss: 0.1600 - accuracy: 0.9421 - auc: 0.9833 - f1_score: 0.9369 - val_loss: 0.1722 - val_accuracy: 0.9343 - val_auc: 0.9814 - val_f1_score: 0.9289\n",
      "Epoch 11/50\n",
      "288/288 [==============================] - 3s 10ms/step - loss: 0.1566 - accuracy: 0.9438 - auc: 0.9838 - f1_score: 0.9390 - val_loss: 0.1724 - val_accuracy: 0.9357 - val_auc: 0.9814 - val_f1_score: 0.9296\n",
      "Epoch 12/50\n",
      "288/288 [==============================] - 3s 9ms/step - loss: 0.1552 - accuracy: 0.9431 - auc: 0.9840 - f1_score: 0.9381 - val_loss: 0.1740 - val_accuracy: 0.9333 - val_auc: 0.9814 - val_f1_score: 0.9265\n",
      "Epoch 13/50\n",
      "288/288 [==============================] - 3s 10ms/step - loss: 0.1479 - accuracy: 0.9475 - auc: 0.9853 - f1_score: 0.9429 - val_loss: 0.1776 - val_accuracy: 0.9355 - val_auc: 0.9804 - val_f1_score: 0.9293\n",
      "Epoch 14/50\n",
      "288/288 [==============================] - 3s 12ms/step - loss: 0.1473 - accuracy: 0.9476 - auc: 0.9854 - f1_score: 0.9430 - val_loss: 0.1768 - val_accuracy: 0.9355 - val_auc: 0.9808 - val_f1_score: 0.9297\n",
      "Epoch 15/50\n",
      "288/288 [==============================] - 3s 9ms/step - loss: 0.1433 - accuracy: 0.9481 - auc: 0.9862 - f1_score: 0.9435 - val_loss: 0.1762 - val_accuracy: 0.9346 - val_auc: 0.9809 - val_f1_score: 0.9290\n",
      "Epoch 16/50\n",
      "288/288 [==============================] - 3s 9ms/step - loss: 0.1389 - accuracy: 0.9493 - auc: 0.9875 - f1_score: 0.9448 - val_loss: 0.1771 - val_accuracy: 0.9369 - val_auc: 0.9806 - val_f1_score: 0.9316\n",
      "Epoch 17/50\n",
      "288/288 [==============================] - 3s 11ms/step - loss: 0.1413 - accuracy: 0.9499 - auc: 0.9864 - f1_score: 0.9454 - val_loss: 0.1794 - val_accuracy: 0.9380 - val_auc: 0.9803 - val_f1_score: 0.9320\n",
      "Epoch 18/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1360 - accuracy: 0.9513 - auc: 0.9870 - f1_score: 0.9470 - val_loss: 0.1791 - val_accuracy: 0.9342 - val_auc: 0.9804 - val_f1_score: 0.9286\n",
      "Epoch 19/50\n",
      "288/288 [==============================] - 2s 5ms/step - loss: 0.1315 - accuracy: 0.9517 - auc: 0.9885 - f1_score: 0.9476 - val_loss: 0.1801 - val_accuracy: 0.9336 - val_auc: 0.9803 - val_f1_score: 0.9275\n",
      "Epoch 20/50\n",
      "288/288 [==============================] - 2s 5ms/step - loss: 0.1306 - accuracy: 0.9534 - auc: 0.9885 - f1_score: 0.9494 - val_loss: 0.1832 - val_accuracy: 0.9357 - val_auc: 0.9797 - val_f1_score: 0.9301\n",
      "Epoch 21/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1240 - accuracy: 0.9539 - auc: 0.9895 - f1_score: 0.9499 - val_loss: 0.1977 - val_accuracy: 0.9333 - val_auc: 0.9784 - val_f1_score: 0.9259\n",
      "Epoch 22/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1265 - accuracy: 0.9525 - auc: 0.9891 - f1_score: 0.9484 - val_loss: 0.1869 - val_accuracy: 0.9364 - val_auc: 0.9802 - val_f1_score: 0.9309\n",
      "Epoch 23/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1213 - accuracy: 0.9552 - auc: 0.9899 - f1_score: 0.9513 - val_loss: 0.1960 - val_accuracy: 0.9354 - val_auc: 0.9795 - val_f1_score: 0.9289\n",
      "Epoch 24/50\n",
      "288/288 [==============================] - 2s 7ms/step - loss: 0.1195 - accuracy: 0.9549 - auc: 0.9899 - f1_score: 0.9509 - val_loss: 0.1953 - val_accuracy: 0.9364 - val_auc: 0.9788 - val_f1_score: 0.9303\n",
      "Epoch 25/50\n",
      "288/288 [==============================] - 2s 5ms/step - loss: 0.1172 - accuracy: 0.9575 - auc: 0.9904 - f1_score: 0.9539 - val_loss: 0.1894 - val_accuracy: 0.9355 - val_auc: 0.9793 - val_f1_score: 0.9301\n",
      "Epoch 26/50\n",
      "288/288 [==============================] - 2s 5ms/step - loss: 0.1184 - accuracy: 0.9581 - auc: 0.9900 - f1_score: 0.9544 - val_loss: 0.1937 - val_accuracy: 0.9347 - val_auc: 0.9792 - val_f1_score: 0.9289\n",
      "Epoch 27/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1120 - accuracy: 0.9589 - auc: 0.9914 - f1_score: 0.9553 - val_loss: 0.2075 - val_accuracy: 0.9368 - val_auc: 0.9784 - val_f1_score: 0.9309\n",
      "Epoch 28/50\n",
      "288/288 [==============================] - 2s 7ms/step - loss: 0.1145 - accuracy: 0.9597 - auc: 0.9906 - f1_score: 0.9561 - val_loss: 0.1965 - val_accuracy: 0.9344 - val_auc: 0.9787 - val_f1_score: 0.9288\n",
      "Epoch 29/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1088 - accuracy: 0.9596 - auc: 0.9918 - f1_score: 0.9561 - val_loss: 0.2174 - val_accuracy: 0.9350 - val_auc: 0.9766 - val_f1_score: 0.9286\n",
      "Epoch 30/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1106 - accuracy: 0.9584 - auc: 0.9916 - f1_score: 0.9548 - val_loss: 0.2116 - val_accuracy: 0.9358 - val_auc: 0.9770 - val_f1_score: 0.9292\n",
      "Epoch 31/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1058 - accuracy: 0.9598 - auc: 0.9922 - f1_score: 0.9562 - val_loss: 0.2072 - val_accuracy: 0.9330 - val_auc: 0.9780 - val_f1_score: 0.9274\n",
      "Epoch 32/50\n",
      "288/288 [==============================] - 2s 5ms/step - loss: 0.1058 - accuracy: 0.9607 - auc: 0.9920 - f1_score: 0.9573 - val_loss: 0.2278 - val_accuracy: 0.9346 - val_auc: 0.9758 - val_f1_score: 0.9276\n",
      "Epoch 33/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1061 - accuracy: 0.9619 - auc: 0.9918 - f1_score: 0.9584 - val_loss: 0.2112 - val_accuracy: 0.9354 - val_auc: 0.9768 - val_f1_score: 0.9297\n",
      "Epoch 34/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1001 - accuracy: 0.9628 - auc: 0.9930 - f1_score: 0.9594 - val_loss: 0.2243 - val_accuracy: 0.9329 - val_auc: 0.9762 - val_f1_score: 0.9272\n",
      "Epoch 35/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.0991 - accuracy: 0.9623 - auc: 0.9930 - f1_score: 0.9589 - val_loss: 0.2262 - val_accuracy: 0.9329 - val_auc: 0.9754 - val_f1_score: 0.9262\n",
      "Epoch 36/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1028 - accuracy: 0.9627 - auc: 0.9926 - f1_score: 0.9594 - val_loss: 0.2270 - val_accuracy: 0.9375 - val_auc: 0.9759 - val_f1_score: 0.9316\n",
      "Epoch 37/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.0952 - accuracy: 0.9648 - auc: 0.9934 - f1_score: 0.9617 - val_loss: 0.2300 - val_accuracy: 0.9316 - val_auc: 0.9761 - val_f1_score: 0.9256\n",
      "Epoch 38/50\n",
      "288/288 [==============================] - 2s 7ms/step - loss: 0.0961 - accuracy: 0.9651 - auc: 0.9933 - f1_score: 0.9620 - val_loss: 0.2247 - val_accuracy: 0.9340 - val_auc: 0.9763 - val_f1_score: 0.9274\n",
      "Epoch 39/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.0949 - accuracy: 0.9641 - auc: 0.9937 - f1_score: 0.9609 - val_loss: 0.2204 - val_accuracy: 0.9340 - val_auc: 0.9768 - val_f1_score: 0.9282\n",
      "Epoch 40/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.0947 - accuracy: 0.9660 - auc: 0.9936 - f1_score: 0.9629 - val_loss: 0.2400 - val_accuracy: 0.9347 - val_auc: 0.9748 - val_f1_score: 0.9281\n",
      "Epoch 41/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.0933 - accuracy: 0.9655 - auc: 0.9938 - f1_score: 0.9624 - val_loss: 0.2639 - val_accuracy: 0.9316 - val_auc: 0.9726 - val_f1_score: 0.9244\n",
      "Epoch 42/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.0938 - accuracy: 0.9653 - auc: 0.9940 - f1_score: 0.9622 - val_loss: 0.2437 - val_accuracy: 0.9340 - val_auc: 0.9736 - val_f1_score: 0.9275\n",
      "Epoch 43/50\n",
      "288/288 [==============================] - 2s 5ms/step - loss: 0.0921 - accuracy: 0.9671 - auc: 0.9940 - f1_score: 0.9641 - val_loss: 0.2411 - val_accuracy: 0.9350 - val_auc: 0.9741 - val_f1_score: 0.9292\n",
      "Epoch 44/50\n",
      "288/288 [==============================] - 2s 5ms/step - loss: 0.0896 - accuracy: 0.9661 - auc: 0.9944 - f1_score: 0.9631 - val_loss: 0.2509 - val_accuracy: 0.9335 - val_auc: 0.9745 - val_f1_score: 0.9279\n",
      "Epoch 45/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.0901 - accuracy: 0.9661 - auc: 0.9941 - f1_score: 0.9631 - val_loss: 0.2558 - val_accuracy: 0.9346 - val_auc: 0.9731 - val_f1_score: 0.9285\n",
      "Epoch 46/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.0866 - accuracy: 0.9676 - auc: 0.9947 - f1_score: 0.9647 - val_loss: 0.2536 - val_accuracy: 0.9316 - val_auc: 0.9728 - val_f1_score: 0.9256\n",
      "Epoch 47/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.0851 - accuracy: 0.9683 - auc: 0.9948 - f1_score: 0.9655 - val_loss: 0.2720 - val_accuracy: 0.9328 - val_auc: 0.9725 - val_f1_score: 0.9261\n",
      "Epoch 48/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.0873 - accuracy: 0.9672 - auc: 0.9944 - f1_score: 0.9643 - val_loss: 0.2617 - val_accuracy: 0.9344 - val_auc: 0.9725 - val_f1_score: 0.9286\n",
      "Epoch 49/50\n",
      "288/288 [==============================] - 2s 5ms/step - loss: 0.0862 - accuracy: 0.9694 - auc: 0.9945 - f1_score: 0.9668 - val_loss: 0.2674 - val_accuracy: 0.9342 - val_auc: 0.9728 - val_f1_score: 0.9280\n",
      "Epoch 50/50\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.0918 - accuracy: 0.9654 - auc: 0.9938 - f1_score: 0.9624 - val_loss: 0.2615 - val_accuracy: 0.9346 - val_auc: 0.9727 - val_f1_score: 0.9281\n",
      "Best epoch: 8\n"
     ]
    }
   ],
   "source": [
    "# Encontre o número ótimo de épocas para treinar o modelo com os hiperparâmetros obtidos na busca\n",
    "\n",
    "# Build the model with the optimal hyperparameters and train it on the data for 50 epochs\n",
    "model_mlp_tn = tuner_mlp.hypermodel.build(best_hps)\n",
    "\n",
    "history_mlp_tn = model_mlp_tn.fit(\n",
    "                                    x=X_train_mlp, \n",
    "                                    y=y_train_mlp, \n",
    "                                    epochs=50, \n",
    "                                    batch_size=BATCH_SIZE, \n",
    "                                    verbose='auto',\n",
    "                                    validation_data=(X_valid_mlp,y_valid_mlp)\n",
    "                                  )\n",
    "\n",
    "val_loss_per_epoch = history_mlp_tn.history['val_loss']\n",
    "best_epoch = val_loss_per_epoch.index(min(val_loss_per_epoch)) + 1\n",
    "print('Best epoch: %d' % (best_epoch,))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2d7733dd",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T19:35:01.389715Z",
     "start_time": "2022-04-01T19:35:00.963878Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 350
    },
    "id": "2d7733dd",
    "outputId": "18bf5833-6258-4d91-e6bf-f10312542fce"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15, 5))\n",
    "plt.subplot(1, 2, 1)\n",
    "plt.plot(history_mlp_tn.history['val_f1_score'], color='blue', label='val_f1_score')\n",
    "plt.plot(history_mlp_tn.history['f1_score'], color='red', label='train_f1_score')\n",
    "plt.title('f1_score')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('f1_score')\n",
    "#plt.ylim(None, 1)\n",
    "plt.legend()\n",
    "plt.subplot(1, 2, 2)\n",
    "plt.plot(history_mlp_tn.history['val_accuracy'], color='blue', label='val_accuracy')\n",
    "plt.plot(history_mlp_tn.history['accuracy'], color='red', label='train_accuracy')\n",
    "plt.title('Accuracy')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('accuracy')\n",
    "#plt.ylim(0, None);\n",
    "plt.legend();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "vriA0x_UleTJ",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T19:35:12.252219Z",
     "start_time": "2022-04-01T19:35:05.656246Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "vriA0x_UleTJ",
    "outputId": "4463dc05-4da7-4f93-d2ab-fc362d568862"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/8\n",
      "288/288 [==============================] - 3s 8ms/step - loss: 0.3071 - accuracy: 0.8724 - auc: 0.9439 - f1_score: 0.8816 - val_loss: 0.1970 - val_accuracy: 0.9264 - val_auc: 0.9763 - val_f1_score: 0.9182\n",
      "Epoch 2/8\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.2122 - accuracy: 0.9259 - auc: 0.9713 - f1_score: 0.9191 - val_loss: 0.1881 - val_accuracy: 0.9287 - val_auc: 0.9782 - val_f1_score: 0.9235\n",
      "Epoch 3/8\n",
      "288/288 [==============================] - 2s 5ms/step - loss: 0.2016 - accuracy: 0.9289 - auc: 0.9737 - f1_score: 0.9228 - val_loss: 0.1819 - val_accuracy: 0.9332 - val_auc: 0.9792 - val_f1_score: 0.9273\n",
      "Epoch 4/8\n",
      "288/288 [==============================] - 2s 7ms/step - loss: 0.1904 - accuracy: 0.9316 - auc: 0.9767 - f1_score: 0.9254 - val_loss: 0.1781 - val_accuracy: 0.9312 - val_auc: 0.9798 - val_f1_score: 0.9249\n",
      "Epoch 5/8\n",
      "288/288 [==============================] - 2s 8ms/step - loss: 0.1853 - accuracy: 0.9315 - auc: 0.9777 - f1_score: 0.9254 - val_loss: 0.1756 - val_accuracy: 0.9312 - val_auc: 0.9805 - val_f1_score: 0.9253\n",
      "Epoch 6/8\n",
      "288/288 [==============================] - 2s 7ms/step - loss: 0.1814 - accuracy: 0.9353 - auc: 0.9787 - f1_score: 0.9295 - val_loss: 0.1762 - val_accuracy: 0.9322 - val_auc: 0.9811 - val_f1_score: 0.9246\n",
      "Epoch 7/8\n",
      "288/288 [==============================] - 2s 5ms/step - loss: 0.1775 - accuracy: 0.9367 - auc: 0.9796 - f1_score: 0.9310 - val_loss: 0.1733 - val_accuracy: 0.9348 - val_auc: 0.9811 - val_f1_score: 0.9289\n",
      "Epoch 8/8\n",
      "288/288 [==============================] - 2s 6ms/step - loss: 0.1706 - accuracy: 0.9378 - auc: 0.9811 - f1_score: 0.9325 - val_loss: 0.1744 - val_accuracy: 0.9332 - val_auc: 0.9809 - val_f1_score: 0.9279\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7fbedcfe3690>"
      ]
     },
     "execution_count": 120,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hypermodel_mlp = tuner_mlp.hypermodel.build(best_hps)\n",
    "\n",
    "# Retrain the model\n",
    "hypermodel_mlp.fit(\n",
    "                    x=X_train_mlp, \n",
    "                    y=y_train_mlp,  \n",
    "                    epochs=best_epoch,\n",
    "                    batch_size=BATCH_SIZE, \n",
    "                    verbose='auto',\n",
    "                    validation_data=(X_valid_mlp,y_valid_mlp)\n",
    "                  )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "97U1qky-mJOn",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T19:35:20.657161Z",
     "start_time": "2022-04-01T19:35:16.415810Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "97U1qky-mJOn",
    "outputId": "e8d3e2ee-b0f2-483a-e48b-8be8e0e6ae4a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "F1 Score: 0.930\n",
      "Accuracy: 0.935\n",
      "AUC: 0.982\n"
     ]
    }
   ],
   "source": [
    "loss_hp, acc_hp, auc_hp, f1_hp = hypermodel_mlp.evaluate(X_test_mlp, y_test_mlp, batch_size=BATCH_SIZE, verbose='auto')\n",
    "\n",
    "print(\"F1 Score: {:5.3f}\".format(f1_hp))\n",
    "print(\"Accuracy: {:5.3f}\".format(acc_hp))\n",
    "print(\"AUC: {:5.3f}\".format(auc_hp))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "legj8lmsSTZQ",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T19:35:26.448488Z",
     "start_time": "2022-04-01T19:35:26.057179Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "legj8lmsSTZQ",
    "outputId": "7ccbaf7a-e86e-4f6f-890f-41204bda5f78"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Clasification report:\n",
      "               precision    recall  f1-score   support\n",
      "\n",
      "       human       0.95      0.93      0.94      3918\n",
      "     natural       0.92      0.94      0.93      3266\n",
      "\n",
      "    accuracy                           0.94      7184\n",
      "   macro avg       0.93      0.94      0.94      7184\n",
      "weighted avg       0.94      0.94      0.94      7184\n",
      "\n",
      "Confusion matrix:\n",
      " [[3635  283]\n",
      " [ 181 3085]]\n"
     ]
    }
   ],
   "source": [
    "y_predicted_hp_mlp = np.reshape((hypermodel_mlp.predict(X_test_mlp) > 0.5).astype('int8'), len(X_test_mlp))\n",
    "\n",
    "print(\"\\nClasification report:\\n\", classification_report(le_mlp.inverse_transform(y_test_mlp), le_mlp.inverse_transform(y_predicted_hp_mlp)))\n",
    "print(\"Confusion matrix:\\n\", confusion_matrix(y_test_mlp, y_predicted_hp_mlp))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "95c0385f",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T19:50:11.755497Z",
     "start_time": "2022-04-01T19:50:11.628482Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 279
    },
    "id": "95c0385f",
    "outputId": "77510533-2cc1-4229-b910-be54a870dbf7"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "mlp_hp_confusion_matrix = pd.crosstab(np.reshape(le_mlp.inverse_transform(y_test_mlp), len(y_test_mlp)), np.reshape(le_mlp.inverse_transform(y_predicted_hp_mlp), len(y_predicted_hp_mlp)), rownames=['Valor Atual'], colnames=['Valor Predito'])\n",
    "\n",
    "sn.heatmap(mlp_hp_confusion_matrix, annot=True, fmt=\"d\", cmap='viridis', linewidths=.5)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "Sq3ZWpxP5XrO",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 332
    },
    "id": "Sq3ZWpxP5XrO",
    "outputId": "1fdc0f8d-de7b-419d-cc88-af666165a52b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.9354120267260579\n",
      "AUC: 0.9817997750580094\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": [
    "roc_curve_function(X_test_mlp, y_test_mlp, y_predicted_hp_mlp, hypermodel_mlp, rn=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "qEVzrPiMQ89p",
   "metadata": {
    "id": "qEVzrPiMQ89p"
   },
   "source": [
    "### Validação Cruzada - Rede Neural"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ZzRlO1stQ8QX",
   "metadata": {
    "id": "ZzRlO1stQ8QX"
   },
   "outputs": [],
   "source": [
    "# K-fold estratificado com k = 10\n",
    "cv = model_selection.StratifiedKFold(n_splits=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "jUCk0vaMRxpI",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "jUCk0vaMRxpI",
    "outputId": "62f430b0-4566-458f-be61-87cd08a5da52"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "518/518 [==============================] - 3s 4ms/step - loss: 0.2853 - accuracy: 0.8847 - auc: 0.9514 - f1_score: 0.8749\n",
      "Epoch 2/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.2160 - accuracy: 0.9272 - auc: 0.9709 - f1_score: 0.9204\n",
      "Epoch 3/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.2049 - accuracy: 0.9313 - auc: 0.9739 - f1_score: 0.9247\n",
      "Epoch 4/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1970 - accuracy: 0.9338 - auc: 0.9760 - f1_score: 0.9276\n",
      "Epoch 5/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1908 - accuracy: 0.9370 - auc: 0.9772 - f1_score: 0.9312\n",
      "Epoch 6/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1841 - accuracy: 0.9374 - auc: 0.9787 - f1_score: 0.9314\n",
      "Epoch 7/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1797 - accuracy: 0.9405 - auc: 0.9796 - f1_score: 0.9348\n",
      "Epoch 8/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1756 - accuracy: 0.9411 - auc: 0.9806 - f1_score: 0.9356\n",
      "Epoch 9/100\n",
      "518/518 [==============================] - 3s 6ms/step - loss: 0.1741 - accuracy: 0.9416 - auc: 0.9809 - f1_score: 0.9361\n",
      "Epoch 10/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1708 - accuracy: 0.9419 - auc: 0.9815 - f1_score: 0.9364\n",
      "Epoch 11/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1685 - accuracy: 0.9435 - auc: 0.9819 - f1_score: 0.9382\n",
      "Epoch 12/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1623 - accuracy: 0.9443 - auc: 0.9831 - f1_score: 0.9391\n",
      "Epoch 13/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1645 - accuracy: 0.9427 - auc: 0.9826 - f1_score: 0.9373\n",
      "Epoch 14/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1585 - accuracy: 0.9455 - auc: 0.9839 - f1_score: 0.9404\n",
      "Epoch 15/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1596 - accuracy: 0.9441 - auc: 0.9838 - f1_score: 0.9388\n",
      "Epoch 16/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1553 - accuracy: 0.9462 - auc: 0.9846 - f1_score: 0.9412\n",
      "Epoch 17/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1545 - accuracy: 0.9474 - auc: 0.9846 - f1_score: 0.9423\n",
      "Epoch 18/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1524 - accuracy: 0.9481 - auc: 0.9850 - f1_score: 0.9432\n",
      "Epoch 19/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1509 - accuracy: 0.9483 - auc: 0.9853 - f1_score: 0.9434\n",
      "Epoch 20/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1485 - accuracy: 0.9480 - auc: 0.9857 - f1_score: 0.9431\n",
      "Epoch 21/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1474 - accuracy: 0.9486 - auc: 0.9860 - f1_score: 0.9438\n",
      "Epoch 22/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1469 - accuracy: 0.9493 - auc: 0.9859 - f1_score: 0.9445\n",
      "Epoch 23/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1436 - accuracy: 0.9505 - auc: 0.9864 - f1_score: 0.9458\n",
      "Epoch 24/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1430 - accuracy: 0.9502 - auc: 0.9869 - f1_score: 0.9455\n",
      "Epoch 25/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1403 - accuracy: 0.9500 - auc: 0.9873 - f1_score: 0.9454\n",
      "Epoch 26/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1427 - accuracy: 0.9500 - auc: 0.9866 - f1_score: 0.9454\n",
      "Epoch 27/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1364 - accuracy: 0.9523 - auc: 0.9880 - f1_score: 0.9477\n",
      "Epoch 28/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1393 - accuracy: 0.9522 - auc: 0.9874 - f1_score: 0.9476\n",
      "Epoch 29/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1393 - accuracy: 0.9514 - auc: 0.9873 - f1_score: 0.9468\n",
      "Epoch 30/100\n",
      "518/518 [==============================] - 3s 5ms/step - loss: 0.1353 - accuracy: 0.9523 - auc: 0.9882 - f1_score: 0.9478\n",
      "Epoch 31/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1338 - accuracy: 0.9534 - auc: 0.9884 - f1_score: 0.9491\n",
      "Epoch 32/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1317 - accuracy: 0.9533 - auc: 0.9885 - f1_score: 0.9489\n",
      "Epoch 33/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1350 - accuracy: 0.9520 - auc: 0.9882 - f1_score: 0.9473\n",
      "Epoch 34/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1315 - accuracy: 0.9525 - auc: 0.9886 - f1_score: 0.9480\n",
      "Epoch 35/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1303 - accuracy: 0.9526 - auc: 0.9888 - f1_score: 0.9480\n",
      "Epoch 36/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1307 - accuracy: 0.9548 - auc: 0.9884 - f1_score: 0.9505\n",
      "Epoch 37/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1269 - accuracy: 0.9548 - auc: 0.9896 - f1_score: 0.9505\n",
      "Epoch 38/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1251 - accuracy: 0.9559 - auc: 0.9897 - f1_score: 0.9516\n",
      "Epoch 39/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1293 - accuracy: 0.9550 - auc: 0.9891 - f1_score: 0.9508\n",
      "Epoch 40/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1289 - accuracy: 0.9546 - auc: 0.9890 - f1_score: 0.9502\n",
      "Epoch 41/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1293 - accuracy: 0.9532 - auc: 0.9891 - f1_score: 0.9488\n",
      "Epoch 42/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1234 - accuracy: 0.9560 - auc: 0.9900 - f1_score: 0.9518\n",
      "Epoch 43/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1284 - accuracy: 0.9548 - auc: 0.9891 - f1_score: 0.9506\n",
      "Epoch 44/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1269 - accuracy: 0.9538 - auc: 0.9895 - f1_score: 0.9493\n",
      "Epoch 45/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1217 - accuracy: 0.9556 - auc: 0.9901 - f1_score: 0.9514\n",
      "Epoch 46/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1197 - accuracy: 0.9577 - auc: 0.9904 - f1_score: 0.9538\n",
      "Epoch 47/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1203 - accuracy: 0.9573 - auc: 0.9904 - f1_score: 0.9532\n",
      "Epoch 48/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1207 - accuracy: 0.9574 - auc: 0.9904 - f1_score: 0.9534\n",
      "Epoch 49/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1199 - accuracy: 0.9578 - auc: 0.9905 - f1_score: 0.9537\n",
      "Epoch 50/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1181 - accuracy: 0.9574 - auc: 0.9908 - f1_score: 0.9533\n",
      "Epoch 51/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1186 - accuracy: 0.9577 - auc: 0.9908 - f1_score: 0.9537\n",
      "Epoch 52/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1162 - accuracy: 0.9585 - auc: 0.9911 - f1_score: 0.9545\n",
      "Epoch 53/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1155 - accuracy: 0.9569 - auc: 0.9912 - f1_score: 0.9528\n",
      "Epoch 54/100\n",
      "518/518 [==============================] - 3s 6ms/step - loss: 0.1147 - accuracy: 0.9580 - auc: 0.9913 - f1_score: 0.9540\n",
      "Epoch 55/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1170 - accuracy: 0.9594 - auc: 0.9909 - f1_score: 0.9556\n",
      "Epoch 56/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1178 - accuracy: 0.9572 - auc: 0.9909 - f1_score: 0.9531\n",
      "Epoch 57/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1151 - accuracy: 0.9586 - auc: 0.9914 - f1_score: 0.9547\n",
      "Epoch 58/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1145 - accuracy: 0.9588 - auc: 0.9911 - f1_score: 0.9548\n",
      "Epoch 59/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1126 - accuracy: 0.9595 - auc: 0.9916 - f1_score: 0.9557\n",
      "Epoch 60/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1148 - accuracy: 0.9588 - auc: 0.9914 - f1_score: 0.9548\n",
      "Epoch 61/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1138 - accuracy: 0.9572 - auc: 0.9917 - f1_score: 0.9530\n",
      "Epoch 62/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1115 - accuracy: 0.9597 - auc: 0.9918 - f1_score: 0.9559\n",
      "Epoch 63/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1121 - accuracy: 0.9610 - auc: 0.9916 - f1_score: 0.9573\n",
      "Epoch 64/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1125 - accuracy: 0.9605 - auc: 0.9915 - f1_score: 0.9567\n",
      "Epoch 65/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1098 - accuracy: 0.9617 - auc: 0.9919 - f1_score: 0.9581\n",
      "Epoch 66/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1115 - accuracy: 0.9596 - auc: 0.9916 - f1_score: 0.9557\n",
      "Epoch 67/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1107 - accuracy: 0.9602 - auc: 0.9918 - f1_score: 0.9565\n",
      "Epoch 68/100\n",
      "518/518 [==============================] - 3s 5ms/step - loss: 0.1112 - accuracy: 0.9602 - auc: 0.9916 - f1_score: 0.9564\n",
      "Epoch 69/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1108 - accuracy: 0.9604 - auc: 0.9918 - f1_score: 0.9567\n",
      "Epoch 70/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1088 - accuracy: 0.9594 - auc: 0.9922 - f1_score: 0.9556\n",
      "Epoch 71/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1126 - accuracy: 0.9594 - auc: 0.9913 - f1_score: 0.9556\n",
      "Epoch 72/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1074 - accuracy: 0.9606 - auc: 0.9922 - f1_score: 0.9569\n",
      "Epoch 73/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1078 - accuracy: 0.9607 - auc: 0.9923 - f1_score: 0.9570\n",
      "Epoch 74/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1092 - accuracy: 0.9601 - auc: 0.9920 - f1_score: 0.9563\n",
      "Epoch 75/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1087 - accuracy: 0.9607 - auc: 0.9922 - f1_score: 0.9570\n",
      "Epoch 76/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1079 - accuracy: 0.9619 - auc: 0.9921 - f1_score: 0.9583\n",
      "Epoch 77/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1067 - accuracy: 0.9625 - auc: 0.9922 - f1_score: 0.9589\n",
      "Epoch 78/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1033 - accuracy: 0.9628 - auc: 0.9929 - f1_score: 0.9593\n",
      "Epoch 79/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1073 - accuracy: 0.9621 - auc: 0.9922 - f1_score: 0.9585\n",
      "Epoch 80/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1080 - accuracy: 0.9614 - auc: 0.9920 - f1_score: 0.9578\n",
      "Epoch 81/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1088 - accuracy: 0.9610 - auc: 0.9921 - f1_score: 0.9573\n",
      "Epoch 82/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1073 - accuracy: 0.9618 - auc: 0.9924 - f1_score: 0.9582\n",
      "Epoch 83/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1049 - accuracy: 0.9630 - auc: 0.9925 - f1_score: 0.9596\n",
      "Score for fold 1: loss of 0.2875346839427948; accuracy of 0.931106448173523; auc of 0.9695867896080017; f1_score of 0.9244275093078613\n",
      "Epoch 1/100\n",
      "518/518 [==============================] - 3s 4ms/step - loss: 0.2970 - accuracy: 0.8789 - auc: 0.9477 - f1_score: 0.8664\n",
      "Epoch 2/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.2135 - accuracy: 0.9252 - auc: 0.9714 - f1_score: 0.9182\n",
      "Epoch 3/100\n",
      "518/518 [==============================] - 3s 5ms/step - loss: 0.2017 - accuracy: 0.9297 - auc: 0.9746 - f1_score: 0.9231\n",
      "Epoch 4/100\n",
      "518/518 [==============================] - 3s 6ms/step - loss: 0.1938 - accuracy: 0.9338 - auc: 0.9761 - f1_score: 0.9276\n",
      "Epoch 5/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1893 - accuracy: 0.9339 - auc: 0.9774 - f1_score: 0.9277\n",
      "Epoch 6/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1804 - accuracy: 0.9369 - auc: 0.9796 - f1_score: 0.9311\n",
      "Epoch 7/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1784 - accuracy: 0.9396 - auc: 0.9795 - f1_score: 0.9340\n",
      "Epoch 8/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1739 - accuracy: 0.9398 - auc: 0.9806 - f1_score: 0.9342\n",
      "Epoch 9/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1702 - accuracy: 0.9403 - auc: 0.9815 - f1_score: 0.9349\n",
      "Epoch 10/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1690 - accuracy: 0.9401 - auc: 0.9817 - f1_score: 0.9346\n",
      "Epoch 11/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1646 - accuracy: 0.9428 - auc: 0.9828 - f1_score: 0.9375\n",
      "Epoch 12/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1608 - accuracy: 0.9428 - auc: 0.9832 - f1_score: 0.9375\n",
      "Epoch 13/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1594 - accuracy: 0.9439 - auc: 0.9835 - f1_score: 0.9388\n",
      "Epoch 14/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1596 - accuracy: 0.9439 - auc: 0.9834 - f1_score: 0.9388\n",
      "Epoch 15/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1555 - accuracy: 0.9449 - auc: 0.9842 - f1_score: 0.9399\n",
      "Epoch 16/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1546 - accuracy: 0.9460 - auc: 0.9844 - f1_score: 0.9411\n",
      "Epoch 17/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1526 - accuracy: 0.9468 - auc: 0.9847 - f1_score: 0.9419\n",
      "Epoch 18/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1517 - accuracy: 0.9469 - auc: 0.9849 - f1_score: 0.9420\n",
      "Epoch 19/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1495 - accuracy: 0.9469 - auc: 0.9855 - f1_score: 0.9422\n",
      "Epoch 20/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1464 - accuracy: 0.9488 - auc: 0.9860 - f1_score: 0.9440\n",
      "Epoch 21/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1444 - accuracy: 0.9478 - auc: 0.9864 - f1_score: 0.9429\n",
      "Epoch 22/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1433 - accuracy: 0.9499 - auc: 0.9866 - f1_score: 0.9454\n",
      "Epoch 23/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1439 - accuracy: 0.9477 - auc: 0.9864 - f1_score: 0.9430\n",
      "Epoch 24/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1432 - accuracy: 0.9489 - auc: 0.9867 - f1_score: 0.9440\n",
      "Epoch 25/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1400 - accuracy: 0.9495 - auc: 0.9870 - f1_score: 0.9449\n",
      "Epoch 26/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1358 - accuracy: 0.9507 - auc: 0.9879 - f1_score: 0.9462\n",
      "Epoch 27/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1353 - accuracy: 0.9512 - auc: 0.9878 - f1_score: 0.9467\n",
      "Epoch 28/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1369 - accuracy: 0.9500 - auc: 0.9878 - f1_score: 0.9455\n",
      "Epoch 29/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1342 - accuracy: 0.9514 - auc: 0.9881 - f1_score: 0.9469\n",
      "Epoch 30/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1344 - accuracy: 0.9508 - auc: 0.9883 - f1_score: 0.9463\n",
      "Epoch 31/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1318 - accuracy: 0.9529 - auc: 0.9885 - f1_score: 0.9485\n",
      "Epoch 32/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1315 - accuracy: 0.9520 - auc: 0.9885 - f1_score: 0.9477\n",
      "Epoch 33/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1281 - accuracy: 0.9531 - auc: 0.9890 - f1_score: 0.9488\n",
      "Epoch 34/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1281 - accuracy: 0.9546 - auc: 0.9893 - f1_score: 0.9504\n",
      "Epoch 35/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1273 - accuracy: 0.9532 - auc: 0.9891 - f1_score: 0.9491\n",
      "Epoch 36/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1269 - accuracy: 0.9529 - auc: 0.9894 - f1_score: 0.9486\n",
      "Epoch 37/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1283 - accuracy: 0.9546 - auc: 0.9892 - f1_score: 0.9505\n",
      "Epoch 38/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1278 - accuracy: 0.9542 - auc: 0.9891 - f1_score: 0.9501\n",
      "Epoch 39/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1248 - accuracy: 0.9542 - auc: 0.9896 - f1_score: 0.9501\n",
      "Epoch 40/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1239 - accuracy: 0.9542 - auc: 0.9899 - f1_score: 0.9501\n",
      "Epoch 41/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1250 - accuracy: 0.9548 - auc: 0.9896 - f1_score: 0.9506\n",
      "Epoch 42/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1236 - accuracy: 0.9560 - auc: 0.9898 - f1_score: 0.9520\n",
      "Epoch 43/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1191 - accuracy: 0.9560 - auc: 0.9904 - f1_score: 0.9521\n",
      "Epoch 44/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1222 - accuracy: 0.9560 - auc: 0.9898 - f1_score: 0.9522\n",
      "Epoch 45/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1233 - accuracy: 0.9552 - auc: 0.9899 - f1_score: 0.9512\n",
      "Epoch 46/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1201 - accuracy: 0.9553 - auc: 0.9904 - f1_score: 0.9513\n",
      "Epoch 47/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1187 - accuracy: 0.9555 - auc: 0.9905 - f1_score: 0.9516\n",
      "Epoch 48/100\n",
      "518/518 [==============================] - 3s 6ms/step - loss: 0.1187 - accuracy: 0.9563 - auc: 0.9906 - f1_score: 0.9524\n",
      "Epoch 49/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1163 - accuracy: 0.9583 - auc: 0.9908 - f1_score: 0.9545\n",
      "Epoch 50/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1163 - accuracy: 0.9578 - auc: 0.9908 - f1_score: 0.9540\n",
      "Epoch 51/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1150 - accuracy: 0.9557 - auc: 0.9909 - f1_score: 0.9517\n",
      "Epoch 52/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1179 - accuracy: 0.9561 - auc: 0.9907 - f1_score: 0.9522\n",
      "Epoch 53/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1148 - accuracy: 0.9563 - auc: 0.9910 - f1_score: 0.9524\n",
      "Epoch 54/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1168 - accuracy: 0.9566 - auc: 0.9907 - f1_score: 0.9527\n",
      "Epoch 55/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1102 - accuracy: 0.9590 - auc: 0.9917 - f1_score: 0.9554\n",
      "Epoch 56/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1137 - accuracy: 0.9582 - auc: 0.9913 - f1_score: 0.9545\n",
      "Epoch 57/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1128 - accuracy: 0.9597 - auc: 0.9915 - f1_score: 0.9562\n",
      "Epoch 58/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1137 - accuracy: 0.9572 - auc: 0.9912 - f1_score: 0.9535\n",
      "Epoch 59/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1112 - accuracy: 0.9587 - auc: 0.9916 - f1_score: 0.9551\n",
      "Epoch 60/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1119 - accuracy: 0.9590 - auc: 0.9916 - f1_score: 0.9553\n",
      "Score for fold 2: loss of 0.21791355311870575; accuracy of 0.9429366588592529; auc of 0.9749271273612976; f1_score of 0.9374523162841797\n",
      "Epoch 1/100\n",
      "518/518 [==============================] - 3s 4ms/step - loss: 0.2832 - accuracy: 0.8888 - auc: 0.9524 - f1_score: 0.8775\n",
      "Epoch 2/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.2095 - accuracy: 0.9257 - auc: 0.9721 - f1_score: 0.9187\n",
      "Epoch 3/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1994 - accuracy: 0.9299 - auc: 0.9746 - f1_score: 0.9234\n",
      "Epoch 4/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1910 - accuracy: 0.9326 - auc: 0.9771 - f1_score: 0.9264\n",
      "Epoch 5/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1869 - accuracy: 0.9333 - auc: 0.9781 - f1_score: 0.9273\n",
      "Epoch 6/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1837 - accuracy: 0.9359 - auc: 0.9787 - f1_score: 0.9301\n",
      "Epoch 7/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1762 - accuracy: 0.9383 - auc: 0.9800 - f1_score: 0.9327\n",
      "Epoch 8/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1731 - accuracy: 0.9380 - auc: 0.9809 - f1_score: 0.9324\n",
      "Epoch 9/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1708 - accuracy: 0.9393 - auc: 0.9815 - f1_score: 0.9337\n",
      "Epoch 10/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1676 - accuracy: 0.9405 - auc: 0.9821 - f1_score: 0.9351\n",
      "Epoch 11/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1662 - accuracy: 0.9410 - auc: 0.9824 - f1_score: 0.9357\n",
      "Epoch 12/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1614 - accuracy: 0.9429 - auc: 0.9831 - f1_score: 0.9376\n",
      "Epoch 13/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1587 - accuracy: 0.9433 - auc: 0.9838 - f1_score: 0.9381\n",
      "Epoch 14/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1573 - accuracy: 0.9456 - auc: 0.9838 - f1_score: 0.9407\n",
      "Epoch 15/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1537 - accuracy: 0.9451 - auc: 0.9847 - f1_score: 0.9401\n",
      "Epoch 16/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1558 - accuracy: 0.9456 - auc: 0.9840 - f1_score: 0.9407\n",
      "Epoch 17/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1515 - accuracy: 0.9461 - auc: 0.9848 - f1_score: 0.9414\n",
      "Epoch 18/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1499 - accuracy: 0.9464 - auc: 0.9855 - f1_score: 0.9416\n",
      "Epoch 19/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1492 - accuracy: 0.9476 - auc: 0.9855 - f1_score: 0.9429\n",
      "Epoch 20/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1459 - accuracy: 0.9469 - auc: 0.9860 - f1_score: 0.9421\n",
      "Epoch 21/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1470 - accuracy: 0.9479 - auc: 0.9859 - f1_score: 0.9433\n",
      "Epoch 22/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1450 - accuracy: 0.9484 - auc: 0.9862 - f1_score: 0.9437\n",
      "Epoch 23/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1421 - accuracy: 0.9485 - auc: 0.9866 - f1_score: 0.9439\n",
      "Epoch 24/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1399 - accuracy: 0.9493 - auc: 0.9873 - f1_score: 0.9447\n",
      "Epoch 25/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1396 - accuracy: 0.9493 - auc: 0.9872 - f1_score: 0.9446\n",
      "Epoch 26/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1360 - accuracy: 0.9501 - auc: 0.9877 - f1_score: 0.9457\n",
      "Epoch 27/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1379 - accuracy: 0.9508 - auc: 0.9873 - f1_score: 0.9465\n",
      "Epoch 28/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1362 - accuracy: 0.9508 - auc: 0.9879 - f1_score: 0.9464\n",
      "Epoch 29/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1350 - accuracy: 0.9512 - auc: 0.9880 - f1_score: 0.9469\n",
      "Epoch 30/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1341 - accuracy: 0.9518 - auc: 0.9880 - f1_score: 0.9475\n",
      "Epoch 31/100\n",
      "518/518 [==============================] - 3s 6ms/step - loss: 0.1284 - accuracy: 0.9526 - auc: 0.9889 - f1_score: 0.9484\n",
      "Epoch 32/100\n",
      "518/518 [==============================] - 3s 5ms/step - loss: 0.1327 - accuracy: 0.9525 - auc: 0.9882 - f1_score: 0.9483\n",
      "Epoch 33/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1328 - accuracy: 0.9517 - auc: 0.9882 - f1_score: 0.9475\n",
      "Epoch 34/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1281 - accuracy: 0.9533 - auc: 0.9891 - f1_score: 0.9492\n",
      "Epoch 35/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1252 - accuracy: 0.9557 - auc: 0.9894 - f1_score: 0.9518\n",
      "Epoch 36/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1256 - accuracy: 0.9539 - auc: 0.9894 - f1_score: 0.9498\n",
      "Epoch 37/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1279 - accuracy: 0.9529 - auc: 0.9890 - f1_score: 0.9488\n",
      "Epoch 38/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1218 - accuracy: 0.9551 - auc: 0.9898 - f1_score: 0.9512\n",
      "Epoch 39/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1217 - accuracy: 0.9546 - auc: 0.9900 - f1_score: 0.9507\n",
      "Epoch 40/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1233 - accuracy: 0.9549 - auc: 0.9898 - f1_score: 0.9509\n",
      "Epoch 41/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1212 - accuracy: 0.9549 - auc: 0.9902 - f1_score: 0.9510\n",
      "Epoch 42/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1223 - accuracy: 0.9544 - auc: 0.9900 - f1_score: 0.9503\n",
      "Epoch 43/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1203 - accuracy: 0.9550 - auc: 0.9904 - f1_score: 0.9511\n",
      "Epoch 44/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1184 - accuracy: 0.9567 - auc: 0.9904 - f1_score: 0.9529\n",
      "Epoch 45/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1162 - accuracy: 0.9564 - auc: 0.9907 - f1_score: 0.9526\n",
      "Epoch 46/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1171 - accuracy: 0.9549 - auc: 0.9908 - f1_score: 0.9510\n",
      "Epoch 47/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1204 - accuracy: 0.9558 - auc: 0.9902 - f1_score: 0.9520\n",
      "Epoch 48/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1180 - accuracy: 0.9560 - auc: 0.9904 - f1_score: 0.9522\n",
      "Epoch 49/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1173 - accuracy: 0.9564 - auc: 0.9907 - f1_score: 0.9527\n",
      "Epoch 50/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1159 - accuracy: 0.9575 - auc: 0.9909 - f1_score: 0.9539\n",
      "Epoch 51/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1156 - accuracy: 0.9570 - auc: 0.9909 - f1_score: 0.9533\n",
      "Epoch 52/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1152 - accuracy: 0.9573 - auc: 0.9910 - f1_score: 0.9536\n",
      "Epoch 53/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1126 - accuracy: 0.9581 - auc: 0.9916 - f1_score: 0.9545\n",
      "Epoch 54/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1143 - accuracy: 0.9579 - auc: 0.9910 - f1_score: 0.9543\n",
      "Epoch 55/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1143 - accuracy: 0.9580 - auc: 0.9913 - f1_score: 0.9544\n",
      "Epoch 56/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1118 - accuracy: 0.9583 - auc: 0.9915 - f1_score: 0.9546\n",
      "Epoch 57/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1155 - accuracy: 0.9568 - auc: 0.9910 - f1_score: 0.9531\n",
      "Epoch 58/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1121 - accuracy: 0.9594 - auc: 0.9913 - f1_score: 0.9559\n",
      "Epoch 59/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1112 - accuracy: 0.9584 - auc: 0.9917 - f1_score: 0.9548\n",
      "Epoch 60/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1084 - accuracy: 0.9594 - auc: 0.9921 - f1_score: 0.9558\n",
      "Epoch 61/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1110 - accuracy: 0.9594 - auc: 0.9915 - f1_score: 0.9560\n",
      "Epoch 62/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1102 - accuracy: 0.9595 - auc: 0.9918 - f1_score: 0.9559\n",
      "Epoch 63/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1098 - accuracy: 0.9587 - auc: 0.9918 - f1_score: 0.9551\n",
      "Epoch 64/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1097 - accuracy: 0.9590 - auc: 0.9917 - f1_score: 0.9554\n",
      "Epoch 65/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1080 - accuracy: 0.9582 - auc: 0.9920 - f1_score: 0.9546\n",
      "Epoch 66/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1080 - accuracy: 0.9606 - auc: 0.9920 - f1_score: 0.9572\n",
      "Epoch 67/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1101 - accuracy: 0.9582 - auc: 0.9919 - f1_score: 0.9546\n",
      "Epoch 68/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1054 - accuracy: 0.9610 - auc: 0.9923 - f1_score: 0.9576\n",
      "Epoch 69/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1058 - accuracy: 0.9594 - auc: 0.9924 - f1_score: 0.9560\n",
      "Epoch 70/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1068 - accuracy: 0.9589 - auc: 0.9921 - f1_score: 0.9553\n",
      "Epoch 71/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1050 - accuracy: 0.9603 - auc: 0.9924 - f1_score: 0.9569\n",
      "Epoch 72/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1062 - accuracy: 0.9602 - auc: 0.9922 - f1_score: 0.9568\n",
      "Epoch 73/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1031 - accuracy: 0.9606 - auc: 0.9926 - f1_score: 0.9572\n",
      "Epoch 74/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1080 - accuracy: 0.9592 - auc: 0.9922 - f1_score: 0.9558\n",
      "Epoch 75/100\n",
      "518/518 [==============================] - 3s 6ms/step - loss: 0.1045 - accuracy: 0.9610 - auc: 0.9923 - f1_score: 0.9576\n",
      "Epoch 76/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1062 - accuracy: 0.9614 - auc: 0.9923 - f1_score: 0.9581\n",
      "Epoch 77/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1048 - accuracy: 0.9612 - auc: 0.9926 - f1_score: 0.9578\n",
      "Epoch 78/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1053 - accuracy: 0.9596 - auc: 0.9925 - f1_score: 0.9561\n",
      "Score for fold 3: loss of 0.2901983857154846; accuracy of 0.9377174377441406; auc of 0.9722349643707275; f1_score of 0.9315487146377563\n",
      "Epoch 1/100\n",
      "518/518 [==============================] - 3s 4ms/step - loss: 0.2853 - accuracy: 0.8861 - auc: 0.9513 - f1_score: 0.8742\n",
      "Epoch 2/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.2149 - accuracy: 0.9244 - auc: 0.9710 - f1_score: 0.9175\n",
      "Epoch 3/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.2027 - accuracy: 0.9270 - auc: 0.9746 - f1_score: 0.9204\n",
      "Epoch 4/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1971 - accuracy: 0.9294 - auc: 0.9758 - f1_score: 0.9231\n",
      "Epoch 5/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1864 - accuracy: 0.9349 - auc: 0.9780 - f1_score: 0.9290\n",
      "Epoch 6/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1838 - accuracy: 0.9364 - auc: 0.9786 - f1_score: 0.9306\n",
      "Epoch 7/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1773 - accuracy: 0.9376 - auc: 0.9801 - f1_score: 0.9320\n",
      "Epoch 8/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1755 - accuracy: 0.9390 - auc: 0.9804 - f1_score: 0.9333\n",
      "Epoch 9/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1707 - accuracy: 0.9393 - auc: 0.9812 - f1_score: 0.9339\n",
      "Epoch 10/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1676 - accuracy: 0.9411 - auc: 0.9821 - f1_score: 0.9357\n",
      "Epoch 11/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1701 - accuracy: 0.9422 - auc: 0.9814 - f1_score: 0.9368\n",
      "Epoch 12/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1624 - accuracy: 0.9422 - auc: 0.9829 - f1_score: 0.9368\n",
      "Epoch 13/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1611 - accuracy: 0.9427 - auc: 0.9834 - f1_score: 0.9374\n",
      "Epoch 14/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1565 - accuracy: 0.9444 - auc: 0.9844 - f1_score: 0.9393\n",
      "Epoch 15/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1577 - accuracy: 0.9448 - auc: 0.9838 - f1_score: 0.9397\n",
      "Epoch 16/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1556 - accuracy: 0.9449 - auc: 0.9843 - f1_score: 0.9400\n",
      "Epoch 17/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1488 - accuracy: 0.9457 - auc: 0.9856 - f1_score: 0.9407\n",
      "Epoch 18/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1484 - accuracy: 0.9451 - auc: 0.9857 - f1_score: 0.9401\n",
      "Epoch 19/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1494 - accuracy: 0.9484 - auc: 0.9857 - f1_score: 0.9436\n",
      "Epoch 20/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1448 - accuracy: 0.9474 - auc: 0.9864 - f1_score: 0.9425\n",
      "Epoch 21/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1460 - accuracy: 0.9462 - auc: 0.9863 - f1_score: 0.9412\n",
      "Epoch 22/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1425 - accuracy: 0.9485 - auc: 0.9866 - f1_score: 0.9437\n",
      "Epoch 23/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1434 - accuracy: 0.9475 - auc: 0.9868 - f1_score: 0.9426\n",
      "Epoch 24/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1389 - accuracy: 0.9507 - auc: 0.9873 - f1_score: 0.9462\n",
      "Epoch 25/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1389 - accuracy: 0.9502 - auc: 0.9870 - f1_score: 0.9457\n",
      "Epoch 26/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1375 - accuracy: 0.9494 - auc: 0.9876 - f1_score: 0.9447\n",
      "Epoch 27/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1375 - accuracy: 0.9514 - auc: 0.9877 - f1_score: 0.9470\n",
      "Epoch 28/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1313 - accuracy: 0.9519 - auc: 0.9885 - f1_score: 0.9474\n",
      "Epoch 29/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1351 - accuracy: 0.9503 - auc: 0.9882 - f1_score: 0.9457\n",
      "Epoch 30/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1315 - accuracy: 0.9532 - auc: 0.9885 - f1_score: 0.9489\n",
      "Epoch 31/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1316 - accuracy: 0.9524 - auc: 0.9885 - f1_score: 0.9481\n",
      "Epoch 32/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1307 - accuracy: 0.9525 - auc: 0.9885 - f1_score: 0.9482\n",
      "Epoch 33/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1295 - accuracy: 0.9517 - auc: 0.9888 - f1_score: 0.9472\n",
      "Epoch 34/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1283 - accuracy: 0.9527 - auc: 0.9892 - f1_score: 0.9484\n",
      "Epoch 35/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1278 - accuracy: 0.9520 - auc: 0.9892 - f1_score: 0.9477\n",
      "Epoch 36/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1254 - accuracy: 0.9527 - auc: 0.9896 - f1_score: 0.9484\n",
      "Epoch 37/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1250 - accuracy: 0.9524 - auc: 0.9898 - f1_score: 0.9481\n",
      "Epoch 38/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1255 - accuracy: 0.9539 - auc: 0.9895 - f1_score: 0.9497\n",
      "Epoch 39/100\n",
      "518/518 [==============================] - 3s 6ms/step - loss: 0.1258 - accuracy: 0.9531 - auc: 0.9898 - f1_score: 0.9489\n",
      "Epoch 40/100\n",
      "518/518 [==============================] - 3s 5ms/step - loss: 0.1219 - accuracy: 0.9535 - auc: 0.9901 - f1_score: 0.9493\n",
      "Epoch 41/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1237 - accuracy: 0.9540 - auc: 0.9898 - f1_score: 0.9500\n",
      "Epoch 42/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1208 - accuracy: 0.9557 - auc: 0.9903 - f1_score: 0.9517\n",
      "Epoch 43/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1199 - accuracy: 0.9558 - auc: 0.9905 - f1_score: 0.9518\n",
      "Epoch 44/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1195 - accuracy: 0.9553 - auc: 0.9905 - f1_score: 0.9512\n",
      "Epoch 45/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1177 - accuracy: 0.9564 - auc: 0.9906 - f1_score: 0.9525\n",
      "Epoch 46/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1174 - accuracy: 0.9559 - auc: 0.9907 - f1_score: 0.9519\n",
      "Epoch 47/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1168 - accuracy: 0.9555 - auc: 0.9910 - f1_score: 0.9515\n",
      "Epoch 48/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1185 - accuracy: 0.9558 - auc: 0.9905 - f1_score: 0.9518\n",
      "Epoch 49/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1149 - accuracy: 0.9568 - auc: 0.9913 - f1_score: 0.9529\n",
      "Epoch 50/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1189 - accuracy: 0.9562 - auc: 0.9906 - f1_score: 0.9523\n",
      "Epoch 51/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1166 - accuracy: 0.9567 - auc: 0.9908 - f1_score: 0.9528\n",
      "Epoch 52/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1133 - accuracy: 0.9573 - auc: 0.9915 - f1_score: 0.9535\n",
      "Epoch 53/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1164 - accuracy: 0.9562 - auc: 0.9910 - f1_score: 0.9522\n",
      "Epoch 54/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1108 - accuracy: 0.9575 - auc: 0.9919 - f1_score: 0.9537\n",
      "Epoch 55/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1146 - accuracy: 0.9568 - auc: 0.9912 - f1_score: 0.9529\n",
      "Epoch 56/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1117 - accuracy: 0.9573 - auc: 0.9915 - f1_score: 0.9535\n",
      "Epoch 57/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1094 - accuracy: 0.9591 - auc: 0.9920 - f1_score: 0.9554\n",
      "Epoch 58/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1121 - accuracy: 0.9574 - auc: 0.9914 - f1_score: 0.9535\n",
      "Epoch 59/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1120 - accuracy: 0.9589 - auc: 0.9916 - f1_score: 0.9552\n",
      "Epoch 60/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1102 - accuracy: 0.9576 - auc: 0.9918 - f1_score: 0.9537\n",
      "Epoch 61/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1133 - accuracy: 0.9572 - auc: 0.9915 - f1_score: 0.9534\n",
      "Epoch 62/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1100 - accuracy: 0.9581 - auc: 0.9919 - f1_score: 0.9544\n",
      "Score for fold 4: loss of 0.23740088939666748; accuracy of 0.930734395980835; auc of 0.9768529534339905; f1_score of 0.924363374710083\n",
      "Epoch 1/100\n",
      "518/518 [==============================] - 3s 4ms/step - loss: 0.3002 - accuracy: 0.8720 - auc: 0.9466 - f1_score: 0.8577\n",
      "Epoch 2/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.2119 - accuracy: 0.9239 - auc: 0.9720 - f1_score: 0.9172\n",
      "Epoch 3/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.2005 - accuracy: 0.9284 - auc: 0.9747 - f1_score: 0.9218\n",
      "Epoch 4/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1934 - accuracy: 0.9316 - auc: 0.9765 - f1_score: 0.9255\n",
      "Epoch 5/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1862 - accuracy: 0.9353 - auc: 0.9778 - f1_score: 0.9295\n",
      "Epoch 6/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1816 - accuracy: 0.9357 - auc: 0.9792 - f1_score: 0.9299\n",
      "Epoch 7/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1790 - accuracy: 0.9376 - auc: 0.9797 - f1_score: 0.9319\n",
      "Epoch 8/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1741 - accuracy: 0.9390 - auc: 0.9804 - f1_score: 0.9336\n",
      "Epoch 9/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1714 - accuracy: 0.9394 - auc: 0.9810 - f1_score: 0.9339\n",
      "Epoch 10/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1706 - accuracy: 0.9406 - auc: 0.9812 - f1_score: 0.9352\n",
      "Epoch 11/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1682 - accuracy: 0.9407 - auc: 0.9817 - f1_score: 0.9354\n",
      "Epoch 12/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1618 - accuracy: 0.9419 - auc: 0.9830 - f1_score: 0.9366\n",
      "Epoch 13/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1599 - accuracy: 0.9423 - auc: 0.9833 - f1_score: 0.9372\n",
      "Epoch 14/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1573 - accuracy: 0.9437 - auc: 0.9839 - f1_score: 0.9388\n",
      "Epoch 15/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1575 - accuracy: 0.9443 - auc: 0.9837 - f1_score: 0.9394\n",
      "Epoch 16/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1534 - accuracy: 0.9457 - auc: 0.9847 - f1_score: 0.9409\n",
      "Epoch 17/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1496 - accuracy: 0.9469 - auc: 0.9854 - f1_score: 0.9420\n",
      "Epoch 18/100\n",
      "518/518 [==============================] - 3s 5ms/step - loss: 0.1504 - accuracy: 0.9460 - auc: 0.9851 - f1_score: 0.9413\n",
      "Epoch 19/100\n",
      "518/518 [==============================] - 3s 6ms/step - loss: 0.1473 - accuracy: 0.9472 - auc: 0.9856 - f1_score: 0.9424\n",
      "Epoch 20/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1488 - accuracy: 0.9471 - auc: 0.9854 - f1_score: 0.9424\n",
      "Epoch 21/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1444 - accuracy: 0.9488 - auc: 0.9862 - f1_score: 0.9443\n",
      "Epoch 22/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1431 - accuracy: 0.9495 - auc: 0.9864 - f1_score: 0.9450\n",
      "Epoch 23/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1424 - accuracy: 0.9496 - auc: 0.9868 - f1_score: 0.9453\n",
      "Epoch 24/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1407 - accuracy: 0.9499 - auc: 0.9868 - f1_score: 0.9455\n",
      "Epoch 25/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1372 - accuracy: 0.9503 - auc: 0.9875 - f1_score: 0.9459\n",
      "Epoch 26/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1375 - accuracy: 0.9512 - auc: 0.9872 - f1_score: 0.9468\n",
      "Epoch 27/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1344 - accuracy: 0.9519 - auc: 0.9881 - f1_score: 0.9477\n",
      "Epoch 28/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1341 - accuracy: 0.9514 - auc: 0.9881 - f1_score: 0.9470\n",
      "Epoch 29/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1328 - accuracy: 0.9523 - auc: 0.9881 - f1_score: 0.9482\n",
      "Epoch 30/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1312 - accuracy: 0.9532 - auc: 0.9885 - f1_score: 0.9490\n",
      "Epoch 31/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1307 - accuracy: 0.9527 - auc: 0.9885 - f1_score: 0.9485\n",
      "Epoch 32/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1294 - accuracy: 0.9532 - auc: 0.9886 - f1_score: 0.9491\n",
      "Epoch 33/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1305 - accuracy: 0.9525 - auc: 0.9885 - f1_score: 0.9483\n",
      "Epoch 34/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1288 - accuracy: 0.9544 - auc: 0.9888 - f1_score: 0.9504\n",
      "Epoch 35/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1262 - accuracy: 0.9553 - auc: 0.9891 - f1_score: 0.9515\n",
      "Epoch 36/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1279 - accuracy: 0.9538 - auc: 0.9890 - f1_score: 0.9498\n",
      "Epoch 37/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1280 - accuracy: 0.9543 - auc: 0.9890 - f1_score: 0.9503\n",
      "Epoch 38/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1239 - accuracy: 0.9549 - auc: 0.9896 - f1_score: 0.9510\n",
      "Epoch 39/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1222 - accuracy: 0.9565 - auc: 0.9898 - f1_score: 0.9526\n",
      "Epoch 40/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1221 - accuracy: 0.9555 - auc: 0.9898 - f1_score: 0.9516\n",
      "Epoch 41/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1186 - accuracy: 0.9570 - auc: 0.9905 - f1_score: 0.9533\n",
      "Epoch 42/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1212 - accuracy: 0.9559 - auc: 0.9900 - f1_score: 0.9520\n",
      "Epoch 43/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1200 - accuracy: 0.9554 - auc: 0.9900 - f1_score: 0.9515\n",
      "Epoch 44/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1190 - accuracy: 0.9558 - auc: 0.9904 - f1_score: 0.9520\n",
      "Epoch 45/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1168 - accuracy: 0.9581 - auc: 0.9905 - f1_score: 0.9544\n",
      "Epoch 46/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1153 - accuracy: 0.9563 - auc: 0.9911 - f1_score: 0.9525\n",
      "Epoch 47/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1160 - accuracy: 0.9582 - auc: 0.9906 - f1_score: 0.9546\n",
      "Epoch 48/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1160 - accuracy: 0.9573 - auc: 0.9909 - f1_score: 0.9536\n",
      "Epoch 49/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1139 - accuracy: 0.9586 - auc: 0.9912 - f1_score: 0.9550\n",
      "Epoch 50/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1142 - accuracy: 0.9581 - auc: 0.9913 - f1_score: 0.9544\n",
      "Epoch 51/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1156 - accuracy: 0.9578 - auc: 0.9908 - f1_score: 0.9541\n",
      "Epoch 52/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1128 - accuracy: 0.9583 - auc: 0.9914 - f1_score: 0.9547\n",
      "Epoch 53/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1152 - accuracy: 0.9574 - auc: 0.9908 - f1_score: 0.9537\n",
      "Epoch 54/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1118 - accuracy: 0.9587 - auc: 0.9917 - f1_score: 0.9552\n",
      "Epoch 55/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1125 - accuracy: 0.9593 - auc: 0.9915 - f1_score: 0.9557\n",
      "Epoch 56/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1116 - accuracy: 0.9590 - auc: 0.9915 - f1_score: 0.9554\n",
      "Epoch 57/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1107 - accuracy: 0.9605 - auc: 0.9916 - f1_score: 0.9570\n",
      "Epoch 58/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1081 - accuracy: 0.9587 - auc: 0.9920 - f1_score: 0.9550\n",
      "Epoch 59/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1102 - accuracy: 0.9603 - auc: 0.9917 - f1_score: 0.9569\n",
      "Epoch 60/100\n",
      "518/518 [==============================] - 3s 6ms/step - loss: 0.1081 - accuracy: 0.9601 - auc: 0.9920 - f1_score: 0.9567\n",
      "Epoch 61/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1083 - accuracy: 0.9592 - auc: 0.9921 - f1_score: 0.9557\n",
      "Epoch 62/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1092 - accuracy: 0.9597 - auc: 0.9917 - f1_score: 0.9563\n",
      "Epoch 63/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1102 - accuracy: 0.9603 - auc: 0.9915 - f1_score: 0.9569\n",
      "Score for fold 5: loss of 0.22352060675621033; accuracy of 0.9380438327789307; auc of 0.9756754040718079; f1_score of 0.93272864818573\n",
      "Epoch 1/100\n",
      "518/518 [==============================] - 3s 4ms/step - loss: 0.2936 - accuracy: 0.8808 - auc: 0.9489 - f1_score: 0.8683\n",
      "Epoch 2/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.2126 - accuracy: 0.9284 - auc: 0.9717 - f1_score: 0.9214\n",
      "Epoch 3/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1999 - accuracy: 0.9311 - auc: 0.9748 - f1_score: 0.9245\n",
      "Epoch 4/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1933 - accuracy: 0.9342 - auc: 0.9763 - f1_score: 0.9282\n",
      "Epoch 5/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1890 - accuracy: 0.9369 - auc: 0.9772 - f1_score: 0.9311\n",
      "Epoch 6/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1807 - accuracy: 0.9386 - auc: 0.9792 - f1_score: 0.9329\n",
      "Epoch 7/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1798 - accuracy: 0.9385 - auc: 0.9795 - f1_score: 0.9329\n",
      "Epoch 8/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1745 - accuracy: 0.9412 - auc: 0.9806 - f1_score: 0.9358\n",
      "Epoch 9/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1710 - accuracy: 0.9410 - auc: 0.9813 - f1_score: 0.9356\n",
      "Epoch 10/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1687 - accuracy: 0.9417 - auc: 0.9819 - f1_score: 0.9363\n",
      "Epoch 11/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1651 - accuracy: 0.9427 - auc: 0.9826 - f1_score: 0.9374\n",
      "Epoch 12/100\n",
      "518/518 [==============================] - 3s 6ms/step - loss: 0.1618 - accuracy: 0.9453 - auc: 0.9833 - f1_score: 0.9403\n",
      "Epoch 13/100\n",
      "518/518 [==============================] - 3s 5ms/step - loss: 0.1614 - accuracy: 0.9445 - auc: 0.9830 - f1_score: 0.9394\n",
      "Epoch 14/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1585 - accuracy: 0.9452 - auc: 0.9839 - f1_score: 0.9402\n",
      "Epoch 15/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1583 - accuracy: 0.9460 - auc: 0.9839 - f1_score: 0.9411\n",
      "Epoch 16/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1525 - accuracy: 0.9474 - auc: 0.9850 - f1_score: 0.9426\n",
      "Epoch 17/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1517 - accuracy: 0.9490 - auc: 0.9851 - f1_score: 0.9443\n",
      "Epoch 18/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1517 - accuracy: 0.9469 - auc: 0.9853 - f1_score: 0.9420\n",
      "Epoch 19/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1501 - accuracy: 0.9471 - auc: 0.9854 - f1_score: 0.9423\n",
      "Epoch 20/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1458 - accuracy: 0.9479 - auc: 0.9862 - f1_score: 0.9432\n",
      "Epoch 21/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1438 - accuracy: 0.9503 - auc: 0.9864 - f1_score: 0.9459\n",
      "Epoch 22/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1431 - accuracy: 0.9495 - auc: 0.9866 - f1_score: 0.9448\n",
      "Epoch 23/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1439 - accuracy: 0.9497 - auc: 0.9863 - f1_score: 0.9450\n",
      "Epoch 24/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1394 - accuracy: 0.9509 - auc: 0.9875 - f1_score: 0.9464\n",
      "Epoch 25/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1390 - accuracy: 0.9517 - auc: 0.9874 - f1_score: 0.9472\n",
      "Epoch 26/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1409 - accuracy: 0.9506 - auc: 0.9869 - f1_score: 0.9461\n",
      "Epoch 27/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1376 - accuracy: 0.9522 - auc: 0.9876 - f1_score: 0.9478\n",
      "Epoch 28/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1354 - accuracy: 0.9526 - auc: 0.9879 - f1_score: 0.9483\n",
      "Epoch 29/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1345 - accuracy: 0.9521 - auc: 0.9880 - f1_score: 0.9477\n",
      "Epoch 30/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1332 - accuracy: 0.9549 - auc: 0.9882 - f1_score: 0.9508\n",
      "Epoch 31/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1324 - accuracy: 0.9536 - auc: 0.9884 - f1_score: 0.9494\n",
      "Epoch 32/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1322 - accuracy: 0.9530 - auc: 0.9883 - f1_score: 0.9487\n",
      "Epoch 33/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1306 - accuracy: 0.9534 - auc: 0.9888 - f1_score: 0.9491\n",
      "Epoch 34/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1269 - accuracy: 0.9550 - auc: 0.9894 - f1_score: 0.9508\n",
      "Epoch 35/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1267 - accuracy: 0.9556 - auc: 0.9893 - f1_score: 0.9515\n",
      "Epoch 36/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1258 - accuracy: 0.9553 - auc: 0.9894 - f1_score: 0.9513\n",
      "Epoch 37/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1256 - accuracy: 0.9549 - auc: 0.9895 - f1_score: 0.9507\n",
      "Epoch 38/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1244 - accuracy: 0.9559 - auc: 0.9897 - f1_score: 0.9519\n",
      "Epoch 39/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1243 - accuracy: 0.9555 - auc: 0.9897 - f1_score: 0.9514\n",
      "Epoch 40/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1248 - accuracy: 0.9548 - auc: 0.9897 - f1_score: 0.9504\n",
      "Epoch 41/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1222 - accuracy: 0.9554 - auc: 0.9901 - f1_score: 0.9513\n",
      "Epoch 42/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1226 - accuracy: 0.9558 - auc: 0.9900 - f1_score: 0.9518\n",
      "Epoch 43/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1196 - accuracy: 0.9563 - auc: 0.9906 - f1_score: 0.9522\n",
      "Epoch 44/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1210 - accuracy: 0.9562 - auc: 0.9902 - f1_score: 0.9521\n",
      "Epoch 45/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1177 - accuracy: 0.9572 - auc: 0.9909 - f1_score: 0.9533\n",
      "Epoch 46/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1192 - accuracy: 0.9570 - auc: 0.9906 - f1_score: 0.9531\n",
      "Epoch 47/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1221 - accuracy: 0.9565 - auc: 0.9900 - f1_score: 0.9524\n",
      "Epoch 48/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1189 - accuracy: 0.9570 - auc: 0.9906 - f1_score: 0.9530\n",
      "Epoch 49/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1171 - accuracy: 0.9573 - auc: 0.9908 - f1_score: 0.9533\n",
      "Epoch 50/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1175 - accuracy: 0.9582 - auc: 0.9907 - f1_score: 0.9543\n",
      "Epoch 51/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1158 - accuracy: 0.9568 - auc: 0.9910 - f1_score: 0.9528\n",
      "Epoch 52/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1148 - accuracy: 0.9587 - auc: 0.9911 - f1_score: 0.9550\n",
      "Epoch 53/100\n",
      "518/518 [==============================] - 3s 6ms/step - loss: 0.1141 - accuracy: 0.9580 - auc: 0.9913 - f1_score: 0.9541\n",
      "Epoch 54/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1142 - accuracy: 0.9593 - auc: 0.9913 - f1_score: 0.9555\n",
      "Epoch 55/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1149 - accuracy: 0.9585 - auc: 0.9912 - f1_score: 0.9547\n",
      "Epoch 56/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1129 - accuracy: 0.9593 - auc: 0.9915 - f1_score: 0.9554\n",
      "Epoch 57/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1156 - accuracy: 0.9580 - auc: 0.9910 - f1_score: 0.9541\n",
      "Epoch 58/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1103 - accuracy: 0.9603 - auc: 0.9918 - f1_score: 0.9566\n",
      "Epoch 59/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1107 - accuracy: 0.9606 - auc: 0.9918 - f1_score: 0.9569\n",
      "Epoch 60/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1107 - accuracy: 0.9599 - auc: 0.9918 - f1_score: 0.9562\n",
      "Epoch 61/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1100 - accuracy: 0.9605 - auc: 0.9919 - f1_score: 0.9569\n",
      "Epoch 62/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1096 - accuracy: 0.9599 - auc: 0.9917 - f1_score: 0.9562\n",
      "Epoch 63/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1117 - accuracy: 0.9596 - auc: 0.9916 - f1_score: 0.9558\n",
      "Epoch 64/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1080 - accuracy: 0.9607 - auc: 0.9922 - f1_score: 0.9569\n",
      "Epoch 65/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1106 - accuracy: 0.9592 - auc: 0.9918 - f1_score: 0.9553\n",
      "Epoch 66/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1085 - accuracy: 0.9596 - auc: 0.9923 - f1_score: 0.9557\n",
      "Epoch 67/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1074 - accuracy: 0.9604 - auc: 0.9923 - f1_score: 0.9567\n",
      "Epoch 68/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1101 - accuracy: 0.9604 - auc: 0.9920 - f1_score: 0.9567\n",
      "Epoch 69/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1054 - accuracy: 0.9609 - auc: 0.9925 - f1_score: 0.9571\n",
      "Epoch 70/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1116 - accuracy: 0.9596 - auc: 0.9916 - f1_score: 0.9558\n",
      "Epoch 71/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1067 - accuracy: 0.9602 - auc: 0.9923 - f1_score: 0.9565\n",
      "Epoch 72/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1054 - accuracy: 0.9626 - auc: 0.9924 - f1_score: 0.9591\n",
      "Epoch 73/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1074 - accuracy: 0.9604 - auc: 0.9923 - f1_score: 0.9567\n",
      "Epoch 74/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1099 - accuracy: 0.9594 - auc: 0.9919 - f1_score: 0.9555\n",
      "Score for fold 6: loss of 0.2694286108016968; accuracy of 0.9338670372962952; auc of 0.974009096622467; f1_score of 0.9268668293952942\n",
      "Epoch 1/100\n",
      "518/518 [==============================] - 3s 4ms/step - loss: 0.2898 - accuracy: 0.8835 - auc: 0.9503 - f1_score: 0.8713\n",
      "Epoch 2/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.2098 - accuracy: 0.9241 - auc: 0.9724 - f1_score: 0.9176\n",
      "Epoch 3/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1996 - accuracy: 0.9293 - auc: 0.9749 - f1_score: 0.9228\n",
      "Epoch 4/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1910 - accuracy: 0.9339 - auc: 0.9772 - f1_score: 0.9276\n",
      "Epoch 5/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1879 - accuracy: 0.9368 - auc: 0.9777 - f1_score: 0.9309\n",
      "Epoch 6/100\n",
      "518/518 [==============================] - 4s 7ms/step - loss: 0.1819 - accuracy: 0.9374 - auc: 0.9789 - f1_score: 0.9316\n",
      "Epoch 7/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1796 - accuracy: 0.9381 - auc: 0.9797 - f1_score: 0.9324\n",
      "Epoch 8/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1763 - accuracy: 0.9403 - auc: 0.9802 - f1_score: 0.9347\n",
      "Epoch 9/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1704 - accuracy: 0.9415 - auc: 0.9813 - f1_score: 0.9362\n",
      "Epoch 10/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1668 - accuracy: 0.9435 - auc: 0.9821 - f1_score: 0.9382\n",
      "Epoch 11/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1675 - accuracy: 0.9430 - auc: 0.9818 - f1_score: 0.9377\n",
      "Epoch 12/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1633 - accuracy: 0.9434 - auc: 0.9828 - f1_score: 0.9382\n",
      "Epoch 13/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1596 - accuracy: 0.9442 - auc: 0.9836 - f1_score: 0.9391\n",
      "Epoch 14/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1570 - accuracy: 0.9453 - auc: 0.9841 - f1_score: 0.9402\n",
      "Epoch 15/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1573 - accuracy: 0.9452 - auc: 0.9839 - f1_score: 0.9402\n",
      "Epoch 16/100\n",
      "518/518 [==============================] - 2s 4ms/step - loss: 0.1539 - accuracy: 0.9474 - auc: 0.9847 - f1_score: 0.9425\n",
      "Epoch 17/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1529 - accuracy: 0.9483 - auc: 0.9847 - f1_score: 0.9436\n",
      "Epoch 18/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1512 - accuracy: 0.9481 - auc: 0.9852 - f1_score: 0.9433\n",
      "Epoch 19/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1464 - accuracy: 0.9491 - auc: 0.9858 - f1_score: 0.9445\n",
      "Epoch 20/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1467 - accuracy: 0.9502 - auc: 0.9859 - f1_score: 0.9456\n",
      "Epoch 21/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1455 - accuracy: 0.9485 - auc: 0.9862 - f1_score: 0.9437\n",
      "Epoch 22/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1406 - accuracy: 0.9500 - auc: 0.9869 - f1_score: 0.9454\n",
      "Epoch 23/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1415 - accuracy: 0.9503 - auc: 0.9868 - f1_score: 0.9458\n",
      "Epoch 24/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1384 - accuracy: 0.9520 - auc: 0.9872 - f1_score: 0.9476\n",
      "Epoch 25/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1389 - accuracy: 0.9511 - auc: 0.9871 - f1_score: 0.9467\n",
      "Epoch 26/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1373 - accuracy: 0.9514 - auc: 0.9874 - f1_score: 0.9470\n",
      "Epoch 27/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1359 - accuracy: 0.9528 - auc: 0.9879 - f1_score: 0.9484\n",
      "Epoch 28/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1361 - accuracy: 0.9515 - auc: 0.9877 - f1_score: 0.9470\n",
      "Epoch 29/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1328 - accuracy: 0.9529 - auc: 0.9883 - f1_score: 0.9484\n",
      "Epoch 30/100\n",
      "518/518 [==============================] - 2s 5ms/step - loss: 0.1324 - accuracy: 0.9520 - auc: 0.9884 - f1_score: 0.9475\n",
      "Epoch 31/100\n",
      "418/518 [=======================>......] - ETA: 0s - loss: 0.1322 - accuracy: 0.9536 - auc: 0.9882 - f1_score: 0.9495"
     ]
    }
   ],
   "source": [
    "inputs = np.concatenate((X_train_mlp, X_valid_mlp, X_test_mlp), axis=0)\n",
    "targets = np.concatenate((y_train_mlp, y_valid_mlp, y_test_mlp), axis=0)\n",
    "\n",
    "BATCH_SIZE = 50    # Tamanho do batch de treinamento\n",
    "N_EPOCHS = 100\n",
    "\n",
    "fold_no = 1\n",
    "loss_per_fold = []\n",
    "acc_per_fold = []\n",
    "auc_per_fold = []\n",
    "f1_per_fold = []\n",
    "\n",
    "for train_index, test_index in cv.split(inputs, targets):\n",
    "  INPUT_SHAPE = (X_train_mlp.shape[1],)\n",
    "  DROPOUT = 0.3\n",
    "\n",
    "  model_cv = keras.models.Sequential([\n",
    "                                        keras.layers.Dense(100, activation='relu', input_shape=INPUT_SHAPE),\n",
    "                                        keras.layers.Dropout(DROPOUT),\n",
    "                                        keras.layers.Dense(50, activation='relu'),\n",
    "                                        keras.layers.Dropout(DROPOUT),\n",
    "                                        keras.layers.Dense(10, activation='relu'),\n",
    "                                        keras.layers.Dropout(DROPOUT),\n",
    "                                        keras.layers.Dense(1, activation = \"sigmoid\")\n",
    "  ])\n",
    "\n",
    "  f1_score = F1Score(num_classes=2, average=\"micro\", threshold=0.5)\n",
    "\n",
    "  model_cv.compile(loss='binary_crossentropy', optimizer=\"Adam\", metrics= ['accuracy', keras.metrics.AUC(name='auc'), f1_score])\n",
    "\n",
    "  callback_cv = tf.keras.callbacks.EarlyStopping(monitor='loss', patience=5) \n",
    "  # This callback will stop the training when there is no improvement in the loss for five consecutive epochs.\n",
    "\n",
    "  history_cv = model_cv.fit(\n",
    "                              x=inputs[train_index], \n",
    "                              y=targets[train_index], \n",
    "                              epochs=N_EPOCHS, \n",
    "                              batch_size=BATCH_SIZE, \n",
    "                              verbose='auto', \n",
    "                              # validation_data=(X_valid_mlp,y_valid_mlp),\n",
    "                              callbacks=[callback_cv],\n",
    "                            )\n",
    "  # Generate metrics\n",
    "  scores = model_cv.evaluate(inputs[test_index], targets[test_index], verbose=0)\n",
    "\n",
    "  print(f'Score for fold {fold_no}: {model_cv.metrics_names[0]} of {scores[0]}; {model_cv.metrics_names[1]} of {scores[1]}; {model_cv.metrics_names[2]} of {scores[2]}; {model_cv.metrics_names[3]} of {scores[3]}')\n",
    "  \n",
    "  loss_per_fold.append(scores[0])\n",
    "  acc_per_fold.append(scores[1])\n",
    "  auc_per_fold.append(scores[2])\n",
    "  f1_per_fold.append(scores[3])\n",
    "\n",
    "  # Increase fold number\n",
    "  fold_no = fold_no + 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7JrZMgPPcfgN",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "7JrZMgPPcfgN",
    "outputId": "26132a37-2b01-485b-e78c-70ff7a8557c2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:     0.936 +/- 0.003\n",
      "F1_score:     0.930 +/- 0.003\n",
      "Roc_auc:      0.974 +/- 0.003\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Print metrics\n",
    "print(\n",
    "        f\"Accuracy:     {np.mean(acc_per_fold):.3f} +/- {np.std(acc_per_fold):.3f}\\n\"\n",
    "        f\"F1_score:     {np.mean(f1_per_fold):.3f} +/- {np.std(f1_per_fold):.3f}\\n\"\n",
    "        f\"Roc_auc:      {np.mean(auc_per_fold):.3f} +/- {np.std(auc_per_fold):.3f}\\n\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "kIU_GVDxxYaJ",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "kIU_GVDxxYaJ",
    "outputId": "1e071c4c-2710-4712-ab95-ccf2b244f0c1"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy per fold:     [0.9335420727729797, 0.9349339008331299, 0.9380654096603394, 0.930734395980835, 0.9380438327789307, 0.9335189461708069, 0.9349112510681152, 0.9397841691970825, 0.939436137676239, 0.9363034963607788]\n",
      "F1_score per fold:     [0.9269598722457886, 0.9280492067337036, 0.9327794909477234, 0.9254401922225952, 0.9324734210968018, 0.9269598722457886, 0.9289243817329407, 0.9339946508407593, 0.9334353804588318, 0.9306555986404419]\n",
      "Roc_auc per fold:      [0.9742457270622253, 0.972420871257782, 0.9675591588020325, 0.9754006862640381, 0.9762335419654846, 0.9753592014312744, 0.9734725952148438, 0.977726936340332, 0.9736078977584839, 0.9730005264282227]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Print metrics for all folds\n",
    "print(\n",
    "      f\"Accuracy per fold:     {acc_per_fold}\\n\"\n",
    "      f\"F1_score per fold:     {f1_per_fold}\\n\"\n",
    "      f\"Roc_auc per fold:      {auc_per_fold}\\n\"\n",
    ")"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "incendios_alberta_pre_processamento_versao_final.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.11"
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}