Detecção_de_Fraude_em_Cartões_de_Crédito.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Detecção de Fraude em Cartões de Crédito.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/JoaoGustavo29/Portfolio_DataScience/blob/master/Detec%C3%A7%C3%A3o_de_Fraude_em_Cart%C3%B5es_de_Cr%C3%A9dito.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sC2BFMOKN1M7"
      },
      "source": [
        "---\n",
        "<img alt=\"Colaboratory logo\" width=\"15%\" src=\"https://raw.githubusercontent.com/carlosfab/escola-data-science/master/img/novo_logo_bg_claro.png\">\n",
        "\n",
        "#### **Data Science na Prática 2.0**\n",
        "*by [sigmoidal.ai](https://sigmoidal.ai)*  \n",
        "\n",
        "---"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IR56trFcPcC1"
      },
      "source": [
        "# Detecção de Fraudes em Cartões de Crédito\n",
        "\n",
        "Neste projeto, iremos abordar o problema das fraudes em cartões de crédito, uma das principais preocupações das instituições financeiras como bancos e *fintechs*. Apenas no Brasil, cerca de 12,1 milhões de pessoas já foram vítimas de algum tipo de fraude financeira no último ano. Traduzindo em valores, os golpes financeiros ultrapassaram a cifra de R$ 1,8 bilhão de prejuízo por ano para os últimos 12 meses.\n",
        "\n",
        "<p align=center>\n",
        "<img src=\"http://sigmoidal.ai/wp-content/uploads/2019/12/900.jpg\" width=\"60%\"></p>\n",
        "\n",
        "Dentra essas fraudes, aquelas envolvendo cartões de crédito são de grande relevância uma vez que a sua não-detecção acarretará em prejuízos consideráveis, tanto para o consumidor quanto para a instituição financeira.\n",
        "\n",
        "Um outro fator a ser considerado é a quantidade de falsos positivos, ou seja, aquelas vezes em que você tentou fazer uma compra e teve seu cartão bloqueado preventivamente - o que provavelmente gerou estresse e constrangimento.\n",
        "\n",
        "Por todos esses motivos, o investimento na área de detecção de fraudes por meio de Inteligência Artificial vem crescendo a cada ano, representando uma grande oportunidade em *Data Science*. \n",
        "\n",
        "Dispondo de grandes volumes de dados como base histórica, um algoritmo de machine learning apenas um pouco melhor que os anteriores já representa uma economia de milhões de Reais. E esse é o desafio, aprimorar cada vez mais o uso de algoritmos visando inibir ou evitar transações fraudulentas."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0HVmpIxQWT4Y"
      },
      "source": [
        "## **Importando os Dados**\n",
        "\n",
        "Os dados que usaremos neste projeto foram disponibilizados por algumas empresas européias de cartão de crédito. O *dataset* representa as operações financeiras que aconteceram no período de dois dias, onde foram classificadas 492 fraudes em meio a quase 290 mil transações.\n",
        "\n",
        "Como podemos notar, este é um conjunto de dados extremamente desbalanceado, onde as fraudes representam apenas 0,17% do total.\n",
        "\n",
        "Outro detalhe interessante é que as *features* são todas numéricas, e foram descaracterizadas (por problemas ligados à privacidade e segurança). Assim, os nomes das colunas são representados por $[V1, V2, V3 \\dots, V28]$ \n",
        "\n",
        "<p align=center>\n",
        "<img src=\"http://sigmoidal.ai/wp-content/uploads/2019/12/54.jpg\" width=\"50%\"></p>\n",
        "\n",
        "[Na página original dos dados](https://www.kaggle.com/mlg-ulb/creditcardfraud), também é informado que as variáveis passaram por uma transformação conhecida como Análise de Componentes Principais (*Principal Component Analysis* - PCA).\n",
        "\n",
        "A PCA permite a redução da dimensionalidade enquanto mantém o maior número possível de informações. Para conseguir isso, o algoritmo encontra um conjunto novo de recursos - os chamados **componentes**.\n",
        "\n",
        "Esses componentes são em número menor ou igual às variáveis originais. No caso deste projeto, os componentes achados pela transformação da PCA são as próprias colunas $[V1, V2, V3 \\dots, V28]$."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oQ4bNy7udtEE"
      },
      "source": [
        "# importar os pacotes necessários\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "from sklearn.metrics import roc_auc_score, roc_curve, accuracy_score\n",
        " \n",
        "# Importar warnings filter\n",
        "from warnings import simplefilter\n",
        "\n",
        "# Ignorar futuros warnings\n",
        "simplefilter(action='ignore', category=FutureWarning)\n",
        "\n",
        "# Configurar o notebook\n",
        "sns.set_style('dark')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "g5XDTWW5LRVe"
      },
      "source": [
        "Para facilitar a aquisição dos dados, criei um *link* com acesso direto ao `csv`. Com isso, basta importar o conjunto de dados usando o método `pd.read_csv()`."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4NxUOfDOj2j8"
      },
      "source": [
        "# Importar o arquivo\n",
        "file_path = \"https://www.dropbox.com/s/b44o3t3ehmnx2b7/creditcard.csv?dl=1\"\n",
        "\n",
        "# Importar o arquivo para um DataFrame\n",
        "df = pd.read_csv(file_path)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7nF_Dhd81Cvq"
      },
      "source": [
        "Com os dados importados para dentro de uma estrutura *Dataframe* - e não havendo a necessidade de mais nenhum ajuste ou configuração nesta etapa, pode-se iniciar uma análise exploratória dos dados a fim de preparar um modelo de *Machine Learning*."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8UtXyZ6stlSM"
      },
      "source": [
        "## **Análise Exploratória**\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KktE8anbKNRN"
      },
      "source": [
        "A fim de conhecermos nosso DataFrame, vamos checar como são as 5 primeiras entradas.\n",
        "\n",
        "* É possível notar como os daods ficaram após a transformação após o PCA (Análise de Componentes Principais), assim deixando as features descaracterizadas (anônimas) e em valores numéricos. Para um melhor entendimento sobre PCA, [clique aqui](https://jakevdp.github.io/PythonDataScienceHandbook/05.09-principal-component-analysis.html).\n",
        "\n",
        "* Para que a análise inicial não fosse prejudicada, as colunas `Time` e `Amount` foram preservadas com seus valores originais.\n",
        "\n",
        "* Nossa vairiávvel alvo está localizada na coluna `Class`, onde:\n",
        "  * **1** é uma transação fraudulenta;\n",
        "  * **0** é uma transação comum."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 224
        },
        "id": "3-xdFBpNKM4D",
        "outputId": "5d454d28-878c-4148-8f1f-e62e2f60625e"
      },
      "source": [
        "# Visualizar as 5 primeiras entradas\n",
        "df.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Time</th>\n",
              "      <th>V1</th>\n",
              "      <th>V2</th>\n",
              "      <th>V3</th>\n",
              "      <th>V4</th>\n",
              "      <th>V5</th>\n",
              "      <th>V6</th>\n",
              "      <th>V7</th>\n",
              "      <th>V8</th>\n",
              "      <th>V9</th>\n",
              "      <th>V10</th>\n",
              "      <th>V11</th>\n",
              "      <th>V12</th>\n",
              "      <th>V13</th>\n",
              "      <th>V14</th>\n",
              "      <th>V15</th>\n",
              "      <th>V16</th>\n",
              "      <th>V17</th>\n",
              "      <th>V18</th>\n",
              "      <th>V19</th>\n",
              "      <th>V20</th>\n",
              "      <th>V21</th>\n",
              "      <th>V22</th>\n",
              "      <th>V23</th>\n",
              "      <th>V24</th>\n",
              "      <th>V25</th>\n",
              "      <th>V26</th>\n",
              "      <th>V27</th>\n",
              "      <th>V28</th>\n",
              "      <th>Amount</th>\n",
              "      <th>Class</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.0</td>\n",
              "      <td>-1.359807</td>\n",
              "      <td>-0.072781</td>\n",
              "      <td>2.536347</td>\n",
              "      <td>1.378155</td>\n",
              "      <td>-0.338321</td>\n",
              "      <td>0.462388</td>\n",
              "      <td>0.239599</td>\n",
              "      <td>0.098698</td>\n",
              "      <td>0.363787</td>\n",
              "      <td>0.090794</td>\n",
              "      <td>-0.551600</td>\n",
              "      <td>-0.617801</td>\n",
              "      <td>-0.991390</td>\n",
              "      <td>-0.311169</td>\n",
              "      <td>1.468177</td>\n",
              "      <td>-0.470401</td>\n",
              "      <td>0.207971</td>\n",
              "      <td>0.025791</td>\n",
              "      <td>0.403993</td>\n",
              "      <td>0.251412</td>\n",
              "      <td>-0.018307</td>\n",
              "      <td>0.277838</td>\n",
              "      <td>-0.110474</td>\n",
              "      <td>0.066928</td>\n",
              "      <td>0.128539</td>\n",
              "      <td>-0.189115</td>\n",
              "      <td>0.133558</td>\n",
              "      <td>-0.021053</td>\n",
              "      <td>149.62</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.0</td>\n",
              "      <td>1.191857</td>\n",
              "      <td>0.266151</td>\n",
              "      <td>0.166480</td>\n",
              "      <td>0.448154</td>\n",
              "      <td>0.060018</td>\n",
              "      <td>-0.082361</td>\n",
              "      <td>-0.078803</td>\n",
              "      <td>0.085102</td>\n",
              "      <td>-0.255425</td>\n",
              "      <td>-0.166974</td>\n",
              "      <td>1.612727</td>\n",
              "      <td>1.065235</td>\n",
              "      <td>0.489095</td>\n",
              "      <td>-0.143772</td>\n",
              "      <td>0.635558</td>\n",
              "      <td>0.463917</td>\n",
              "      <td>-0.114805</td>\n",
              "      <td>-0.183361</td>\n",
              "      <td>-0.145783</td>\n",
              "      <td>-0.069083</td>\n",
              "      <td>-0.225775</td>\n",
              "      <td>-0.638672</td>\n",
              "      <td>0.101288</td>\n",
              "      <td>-0.339846</td>\n",
              "      <td>0.167170</td>\n",
              "      <td>0.125895</td>\n",
              "      <td>-0.008983</td>\n",
              "      <td>0.014724</td>\n",
              "      <td>2.69</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>1.0</td>\n",
              "      <td>-1.358354</td>\n",
              "      <td>-1.340163</td>\n",
              "      <td>1.773209</td>\n",
              "      <td>0.379780</td>\n",
              "      <td>-0.503198</td>\n",
              "      <td>1.800499</td>\n",
              "      <td>0.791461</td>\n",
              "      <td>0.247676</td>\n",
              "      <td>-1.514654</td>\n",
              "      <td>0.207643</td>\n",
              "      <td>0.624501</td>\n",
              "      <td>0.066084</td>\n",
              "      <td>0.717293</td>\n",
              "      <td>-0.165946</td>\n",
              "      <td>2.345865</td>\n",
              "      <td>-2.890083</td>\n",
              "      <td>1.109969</td>\n",
              "      <td>-0.121359</td>\n",
              "      <td>-2.261857</td>\n",
              "      <td>0.524980</td>\n",
              "      <td>0.247998</td>\n",
              "      <td>0.771679</td>\n",
              "      <td>0.909412</td>\n",
              "      <td>-0.689281</td>\n",
              "      <td>-0.327642</td>\n",
              "      <td>-0.139097</td>\n",
              "      <td>-0.055353</td>\n",
              "      <td>-0.059752</td>\n",
              "      <td>378.66</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>1.0</td>\n",
              "      <td>-0.966272</td>\n",
              "      <td>-0.185226</td>\n",
              "      <td>1.792993</td>\n",
              "      <td>-0.863291</td>\n",
              "      <td>-0.010309</td>\n",
              "      <td>1.247203</td>\n",
              "      <td>0.237609</td>\n",
              "      <td>0.377436</td>\n",
              "      <td>-1.387024</td>\n",
              "      <td>-0.054952</td>\n",
              "      <td>-0.226487</td>\n",
              "      <td>0.178228</td>\n",
              "      <td>0.507757</td>\n",
              "      <td>-0.287924</td>\n",
              "      <td>-0.631418</td>\n",
              "      <td>-1.059647</td>\n",
              "      <td>-0.684093</td>\n",
              "      <td>1.965775</td>\n",
              "      <td>-1.232622</td>\n",
              "      <td>-0.208038</td>\n",
              "      <td>-0.108300</td>\n",
              "      <td>0.005274</td>\n",
              "      <td>-0.190321</td>\n",
              "      <td>-1.175575</td>\n",
              "      <td>0.647376</td>\n",
              "      <td>-0.221929</td>\n",
              "      <td>0.062723</td>\n",
              "      <td>0.061458</td>\n",
              "      <td>123.50</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>2.0</td>\n",
              "      <td>-1.158233</td>\n",
              "      <td>0.877737</td>\n",
              "      <td>1.548718</td>\n",
              "      <td>0.403034</td>\n",
              "      <td>-0.407193</td>\n",
              "      <td>0.095921</td>\n",
              "      <td>0.592941</td>\n",
              "      <td>-0.270533</td>\n",
              "      <td>0.817739</td>\n",
              "      <td>0.753074</td>\n",
              "      <td>-0.822843</td>\n",
              "      <td>0.538196</td>\n",
              "      <td>1.345852</td>\n",
              "      <td>-1.119670</td>\n",
              "      <td>0.175121</td>\n",
              "      <td>-0.451449</td>\n",
              "      <td>-0.237033</td>\n",
              "      <td>-0.038195</td>\n",
              "      <td>0.803487</td>\n",
              "      <td>0.408542</td>\n",
              "      <td>-0.009431</td>\n",
              "      <td>0.798278</td>\n",
              "      <td>-0.137458</td>\n",
              "      <td>0.141267</td>\n",
              "      <td>-0.206010</td>\n",
              "      <td>0.502292</td>\n",
              "      <td>0.219422</td>\n",
              "      <td>0.215153</td>\n",
              "      <td>69.99</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   Time        V1        V2        V3  ...       V27       V28  Amount  Class\n",
              "0   0.0 -1.359807 -0.072781  2.536347  ...  0.133558 -0.021053  149.62      0\n",
              "1   0.0  1.191857  0.266151  0.166480  ... -0.008983  0.014724    2.69      0\n",
              "2   1.0 -1.358354 -1.340163  1.773209  ... -0.055353 -0.059752  378.66      0\n",
              "3   1.0 -0.966272 -0.185226  1.792993  ...  0.062723  0.061458  123.50      0\n",
              "4   2.0 -1.158233  0.877737  1.548718  ...  0.219422  0.215153   69.99      0\n",
              "\n",
              "[5 rows x 31 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-mzqCTRGKpjg"
      },
      "source": [
        "Por meio do método `describe()`, vamos verificar o resumo estatístico, nele é possível confrimar que as variáveis que sofreram a Transformação PCA, não possuem uma discrepância aparente, assim como a coluna `Time`.\n",
        "\n",
        "No entanto, a variável `Amount`, tem um valor médio (de ambas as classes) é 88.34, mediana igual a 22 e o desvio padrão impressionante de 250.12. Seu valor máximo é de 25691.16, isso explica o desvio padrão, uma vez que a maioria das transações são de quantias menores."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 317
        },
        "id": "NJR71vEKKp5H",
        "outputId": "9d2d0f20-d08f-490c-d86c-d11ca21c887c"
      },
      "source": [
        "df.describe()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Time</th>\n",
              "      <th>V1</th>\n",
              "      <th>V2</th>\n",
              "      <th>V3</th>\n",
              "      <th>V4</th>\n",
              "      <th>V5</th>\n",
              "      <th>V6</th>\n",
              "      <th>V7</th>\n",
              "      <th>V8</th>\n",
              "      <th>V9</th>\n",
              "      <th>V10</th>\n",
              "      <th>V11</th>\n",
              "      <th>V12</th>\n",
              "      <th>V13</th>\n",
              "      <th>V14</th>\n",
              "      <th>V15</th>\n",
              "      <th>V16</th>\n",
              "      <th>V17</th>\n",
              "      <th>V18</th>\n",
              "      <th>V19</th>\n",
              "      <th>V20</th>\n",
              "      <th>V21</th>\n",
              "      <th>V22</th>\n",
              "      <th>V23</th>\n",
              "      <th>V24</th>\n",
              "      <th>V25</th>\n",
              "      <th>V26</th>\n",
              "      <th>V27</th>\n",
              "      <th>V28</th>\n",
              "      <th>Amount</th>\n",
              "      <th>Class</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>count</th>\n",
              "      <td>284807.000000</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>2.848070e+05</td>\n",
              "      <td>284807.000000</td>\n",
              "      <td>284807.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean</th>\n",
              "      <td>94813.859575</td>\n",
              "      <td>3.919560e-15</td>\n",
              "      <td>5.688174e-16</td>\n",
              "      <td>-8.769071e-15</td>\n",
              "      <td>2.782312e-15</td>\n",
              "      <td>-1.552563e-15</td>\n",
              "      <td>2.010663e-15</td>\n",
              "      <td>-1.694249e-15</td>\n",
              "      <td>-1.927028e-16</td>\n",
              "      <td>-3.137024e-15</td>\n",
              "      <td>1.768627e-15</td>\n",
              "      <td>9.170318e-16</td>\n",
              "      <td>-1.810658e-15</td>\n",
              "      <td>1.693438e-15</td>\n",
              "      <td>1.479045e-15</td>\n",
              "      <td>3.482336e-15</td>\n",
              "      <td>1.392007e-15</td>\n",
              "      <td>-7.528491e-16</td>\n",
              "      <td>4.328772e-16</td>\n",
              "      <td>9.049732e-16</td>\n",
              "      <td>5.085503e-16</td>\n",
              "      <td>1.537294e-16</td>\n",
              "      <td>7.959909e-16</td>\n",
              "      <td>5.367590e-16</td>\n",
              "      <td>4.458112e-15</td>\n",
              "      <td>1.453003e-15</td>\n",
              "      <td>1.699104e-15</td>\n",
              "      <td>-3.660161e-16</td>\n",
              "      <td>-1.206049e-16</td>\n",
              "      <td>88.349619</td>\n",
              "      <td>0.001727</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>std</th>\n",
              "      <td>47488.145955</td>\n",
              "      <td>1.958696e+00</td>\n",
              "      <td>1.651309e+00</td>\n",
              "      <td>1.516255e+00</td>\n",
              "      <td>1.415869e+00</td>\n",
              "      <td>1.380247e+00</td>\n",
              "      <td>1.332271e+00</td>\n",
              "      <td>1.237094e+00</td>\n",
              "      <td>1.194353e+00</td>\n",
              "      <td>1.098632e+00</td>\n",
              "      <td>1.088850e+00</td>\n",
              "      <td>1.020713e+00</td>\n",
              "      <td>9.992014e-01</td>\n",
              "      <td>9.952742e-01</td>\n",
              "      <td>9.585956e-01</td>\n",
              "      <td>9.153160e-01</td>\n",
              "      <td>8.762529e-01</td>\n",
              "      <td>8.493371e-01</td>\n",
              "      <td>8.381762e-01</td>\n",
              "      <td>8.140405e-01</td>\n",
              "      <td>7.709250e-01</td>\n",
              "      <td>7.345240e-01</td>\n",
              "      <td>7.257016e-01</td>\n",
              "      <td>6.244603e-01</td>\n",
              "      <td>6.056471e-01</td>\n",
              "      <td>5.212781e-01</td>\n",
              "      <td>4.822270e-01</td>\n",
              "      <td>4.036325e-01</td>\n",
              "      <td>3.300833e-01</td>\n",
              "      <td>250.120109</td>\n",
              "      <td>0.041527</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>min</th>\n",
              "      <td>0.000000</td>\n",
              "      <td>-5.640751e+01</td>\n",
              "      <td>-7.271573e+01</td>\n",
              "      <td>-4.832559e+01</td>\n",
              "      <td>-5.683171e+00</td>\n",
              "      <td>-1.137433e+02</td>\n",
              "      <td>-2.616051e+01</td>\n",
              "      <td>-4.355724e+01</td>\n",
              "      <td>-7.321672e+01</td>\n",
              "      <td>-1.343407e+01</td>\n",
              "      <td>-2.458826e+01</td>\n",
              "      <td>-4.797473e+00</td>\n",
              "      <td>-1.868371e+01</td>\n",
              "      <td>-5.791881e+00</td>\n",
              "      <td>-1.921433e+01</td>\n",
              "      <td>-4.498945e+00</td>\n",
              "      <td>-1.412985e+01</td>\n",
              "      <td>-2.516280e+01</td>\n",
              "      <td>-9.498746e+00</td>\n",
              "      <td>-7.213527e+00</td>\n",
              "      <td>-5.449772e+01</td>\n",
              "      <td>-3.483038e+01</td>\n",
              "      <td>-1.093314e+01</td>\n",
              "      <td>-4.480774e+01</td>\n",
              "      <td>-2.836627e+00</td>\n",
              "      <td>-1.029540e+01</td>\n",
              "      <td>-2.604551e+00</td>\n",
              "      <td>-2.256568e+01</td>\n",
              "      <td>-1.543008e+01</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>25%</th>\n",
              "      <td>54201.500000</td>\n",
              "      <td>-9.203734e-01</td>\n",
              "      <td>-5.985499e-01</td>\n",
              "      <td>-8.903648e-01</td>\n",
              "      <td>-8.486401e-01</td>\n",
              "      <td>-6.915971e-01</td>\n",
              "      <td>-7.682956e-01</td>\n",
              "      <td>-5.540759e-01</td>\n",
              "      <td>-2.086297e-01</td>\n",
              "      <td>-6.430976e-01</td>\n",
              "      <td>-5.354257e-01</td>\n",
              "      <td>-7.624942e-01</td>\n",
              "      <td>-4.055715e-01</td>\n",
              "      <td>-6.485393e-01</td>\n",
              "      <td>-4.255740e-01</td>\n",
              "      <td>-5.828843e-01</td>\n",
              "      <td>-4.680368e-01</td>\n",
              "      <td>-4.837483e-01</td>\n",
              "      <td>-4.988498e-01</td>\n",
              "      <td>-4.562989e-01</td>\n",
              "      <td>-2.117214e-01</td>\n",
              "      <td>-2.283949e-01</td>\n",
              "      <td>-5.423504e-01</td>\n",
              "      <td>-1.618463e-01</td>\n",
              "      <td>-3.545861e-01</td>\n",
              "      <td>-3.171451e-01</td>\n",
              "      <td>-3.269839e-01</td>\n",
              "      <td>-7.083953e-02</td>\n",
              "      <td>-5.295979e-02</td>\n",
              "      <td>5.600000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>50%</th>\n",
              "      <td>84692.000000</td>\n",
              "      <td>1.810880e-02</td>\n",
              "      <td>6.548556e-02</td>\n",
              "      <td>1.798463e-01</td>\n",
              "      <td>-1.984653e-02</td>\n",
              "      <td>-5.433583e-02</td>\n",
              "      <td>-2.741871e-01</td>\n",
              "      <td>4.010308e-02</td>\n",
              "      <td>2.235804e-02</td>\n",
              "      <td>-5.142873e-02</td>\n",
              "      <td>-9.291738e-02</td>\n",
              "      <td>-3.275735e-02</td>\n",
              "      <td>1.400326e-01</td>\n",
              "      <td>-1.356806e-02</td>\n",
              "      <td>5.060132e-02</td>\n",
              "      <td>4.807155e-02</td>\n",
              "      <td>6.641332e-02</td>\n",
              "      <td>-6.567575e-02</td>\n",
              "      <td>-3.636312e-03</td>\n",
              "      <td>3.734823e-03</td>\n",
              "      <td>-6.248109e-02</td>\n",
              "      <td>-2.945017e-02</td>\n",
              "      <td>6.781943e-03</td>\n",
              "      <td>-1.119293e-02</td>\n",
              "      <td>4.097606e-02</td>\n",
              "      <td>1.659350e-02</td>\n",
              "      <td>-5.213911e-02</td>\n",
              "      <td>1.342146e-03</td>\n",
              "      <td>1.124383e-02</td>\n",
              "      <td>22.000000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>75%</th>\n",
              "      <td>139320.500000</td>\n",
              "      <td>1.315642e+00</td>\n",
              "      <td>8.037239e-01</td>\n",
              "      <td>1.027196e+00</td>\n",
              "      <td>7.433413e-01</td>\n",
              "      <td>6.119264e-01</td>\n",
              "      <td>3.985649e-01</td>\n",
              "      <td>5.704361e-01</td>\n",
              "      <td>3.273459e-01</td>\n",
              "      <td>5.971390e-01</td>\n",
              "      <td>4.539234e-01</td>\n",
              "      <td>7.395934e-01</td>\n",
              "      <td>6.182380e-01</td>\n",
              "      <td>6.625050e-01</td>\n",
              "      <td>4.931498e-01</td>\n",
              "      <td>6.488208e-01</td>\n",
              "      <td>5.232963e-01</td>\n",
              "      <td>3.996750e-01</td>\n",
              "      <td>5.008067e-01</td>\n",
              "      <td>4.589494e-01</td>\n",
              "      <td>1.330408e-01</td>\n",
              "      <td>1.863772e-01</td>\n",
              "      <td>5.285536e-01</td>\n",
              "      <td>1.476421e-01</td>\n",
              "      <td>4.395266e-01</td>\n",
              "      <td>3.507156e-01</td>\n",
              "      <td>2.409522e-01</td>\n",
              "      <td>9.104512e-02</td>\n",
              "      <td>7.827995e-02</td>\n",
              "      <td>77.165000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>max</th>\n",
              "      <td>172792.000000</td>\n",
              "      <td>2.454930e+00</td>\n",
              "      <td>2.205773e+01</td>\n",
              "      <td>9.382558e+00</td>\n",
              "      <td>1.687534e+01</td>\n",
              "      <td>3.480167e+01</td>\n",
              "      <td>7.330163e+01</td>\n",
              "      <td>1.205895e+02</td>\n",
              "      <td>2.000721e+01</td>\n",
              "      <td>1.559499e+01</td>\n",
              "      <td>2.374514e+01</td>\n",
              "      <td>1.201891e+01</td>\n",
              "      <td>7.848392e+00</td>\n",
              "      <td>7.126883e+00</td>\n",
              "      <td>1.052677e+01</td>\n",
              "      <td>8.877742e+00</td>\n",
              "      <td>1.731511e+01</td>\n",
              "      <td>9.253526e+00</td>\n",
              "      <td>5.041069e+00</td>\n",
              "      <td>5.591971e+00</td>\n",
              "      <td>3.942090e+01</td>\n",
              "      <td>2.720284e+01</td>\n",
              "      <td>1.050309e+01</td>\n",
              "      <td>2.252841e+01</td>\n",
              "      <td>4.584549e+00</td>\n",
              "      <td>7.519589e+00</td>\n",
              "      <td>3.517346e+00</td>\n",
              "      <td>3.161220e+01</td>\n",
              "      <td>3.384781e+01</td>\n",
              "      <td>25691.160000</td>\n",
              "      <td>1.000000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                Time            V1  ...         Amount          Class\n",
              "count  284807.000000  2.848070e+05  ...  284807.000000  284807.000000\n",
              "mean    94813.859575  3.919560e-15  ...      88.349619       0.001727\n",
              "std     47488.145955  1.958696e+00  ...     250.120109       0.041527\n",
              "min         0.000000 -5.640751e+01  ...       0.000000       0.000000\n",
              "25%     54201.500000 -9.203734e-01  ...       5.600000       0.000000\n",
              "50%     84692.000000  1.810880e-02  ...      22.000000       0.000000\n",
              "75%    139320.500000  1.315642e+00  ...      77.165000       0.000000\n",
              "max    172792.000000  2.454930e+00  ...   25691.160000       1.000000\n",
              "\n",
              "[8 rows x 31 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xAKB6TJcLevo"
      },
      "source": [
        "Quando procuramos saber a respeito dos valores ausentes do dataset, não encontramos nenhum, assim não é nescessário um trabalho de limpeza. Abaixo é possivel verificar que nenhuma coluna possui valores nulos."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "MP8qbx4qLeh5",
        "outputId": "0ce2c129-4d23-4be4-9155-1afc426f0ff2"
      },
      "source": [
        "(df.isnull().sum() / df.shape[0]).sort_values(ascending=False)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Class     0.0\n",
              "V14       0.0\n",
              "V1        0.0\n",
              "V2        0.0\n",
              "V3        0.0\n",
              "V4        0.0\n",
              "V5        0.0\n",
              "V6        0.0\n",
              "V7        0.0\n",
              "V8        0.0\n",
              "V9        0.0\n",
              "V10       0.0\n",
              "V11       0.0\n",
              "V12       0.0\n",
              "V13       0.0\n",
              "V15       0.0\n",
              "Amount    0.0\n",
              "V16       0.0\n",
              "V17       0.0\n",
              "V18       0.0\n",
              "V19       0.0\n",
              "V20       0.0\n",
              "V21       0.0\n",
              "V22       0.0\n",
              "V23       0.0\n",
              "V24       0.0\n",
              "V25       0.0\n",
              "V26       0.0\n",
              "V27       0.0\n",
              "V28       0.0\n",
              "Time      0.0\n",
              "dtype: float64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_XnMLkmyL4ZZ"
      },
      "source": [
        "Na [descrição dos dados](https://www.kaggle.com/mlg-ulb/creditcardfraud) é informado que as entradas relativas à transações fraudulentas representam apenas 0,17% do total.\n",
        "\n",
        "Vamos verificar isso de forma visual, para isso iremos plotar um gráfico de barras.\n",
        "\n",
        "No gráfico abaixo, fica nítida a discrepância entre os dados. Portanto, será nescessário fazer um balanceamento dos dados, para que nosso modelo não seja prejudicado ao treinar com dados desbalanceados.\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 397
        },
        "id": "fcSvLMzjL8ea",
        "outputId": "b4c122f1-02bd-4af4-cc18-0f54c7e521ad"
      },
      "source": [
        "# Checar o balanceamento das classes\n",
        "print(df.Class.value_counts())\n",
        "print('\\nFraudes representam {:.4f}% do dataset.\\n'.format((df[df.Class == 1].shape[0] / df.shape[0]) * 100))\n",
        "\n",
        "# Plotar gráfico de barras para as Classes\n",
        "fig , ax = plt.subplots()\n",
        "sns.countplot('Class', data=df, ax=ax);\n",
        "ax.set_title('Distribuição das Classes')\n",
        "ax.set_ylabel('Quantidade de ocorrências')\n",
        "ax.set_xlabel('Classe')\n",
        "plt.show();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0    284315\n",
            "1       492\n",
            "Name: Class, dtype: int64\n",
            "\n",
            "Fraudes representam 0.1727% do dataset.\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JgBdEmuBNUbr"
      },
      "source": [
        "Para compararmos o comportamento e a distribução das classes, foram plotados 4 gráficos.\n",
        "\n",
        "Vamos ver o que os gráficos demonstram:\n",
        "\n",
        "* 2 gráficos (1 e 3) - Que usam como a referência a dimensão **tempo** (`Time`), no entanto, não foi indentificada nenhum informação a partir desses gráficos.\n",
        "\n",
        "* 2 gráficos (2 e 4) - Que usam como referência o valor das transações (`Amount`), nesses dois foi possível notar que o valor das transações comuns ficam entre 0 e 5000, diferentemente das transações fraudulentas que pouquissimas tem seu valor acima de 500."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 729
        },
        "id": "T9HguEfaNU51",
        "outputId": "656102bb-d688-4b50-a5b4-d3521d78d095"
      },
      "source": [
        "# Plotar gráfico \n",
        "fig, ax = plt.subplots(nrows=4, ncols=1, figsize=(15,10))\n",
        "\n",
        "# Definir número de barras dos gráficos\n",
        "num_bins_time = 45\n",
        "num_bins_amount = 5\n",
        "\n",
        "# Gráfico 1\n",
        "ax[0].hist(df.Time[df.Class == 0], bins=num_bins_time)\n",
        "ax[0].set_title('Transações Legítimas por segundo')\n",
        "ax[0].set_ylabel('Transações')\n",
        "ax[0].set_xlabel('Tempo (Segundos)')\n",
        "\n",
        "# Gráfico 2\n",
        "ax[1].hist(df.Amount[df.Class == 0], bins=num_bins_amount)\n",
        "ax[1].set_title('Transações Legítimas baseadas no valor')\n",
        "ax[1].set_ylabel('Transações')\n",
        "ax[1].set_xlabel('Valor da Transação')\n",
        "\n",
        "# Gráfico 3\n",
        "ax[2].hist(df.Time[df.Class == 1], bins=num_bins_time)\n",
        "ax[2].set_title('Transações Fraudulentas por segundo')\n",
        "ax[2].set_ylabel('Transações')\n",
        "ax[2].set_xlabel('Tempo (Segundos)')\n",
        "\n",
        "# Gráfico 4\n",
        "ax[3].hist(df.Amount[df.Class == 1], bins=num_bins_amount)\n",
        "ax[3].set_title('Transações Fraudulentas baseadas no valor')\n",
        "ax[3].set_ylabel('Transações')\n",
        "ax[3].set_xlabel('Valor da Transação')\n",
        "\n",
        "# Mostrar gráficos\n",
        "plt.tight_layout();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x720 with 4 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EDfBFQfg4tk9"
      },
      "source": [
        "Foram plotados também, boxplots, uma ótima ferramenta para se há diferença no padrão das transações em relação ao seu valor (`Amount`).\n",
        "\n",
        "Portanto, é possivel notar que há uma distribuição diferente para as duas classes, o que deve beneficiar o treinamento dos nossos modelos."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tFzzZm1a4t-n",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 873
        },
        "outputId": "63c475e7-f7fb-47a7-aaad-c32131fafdfb"
      },
      "source": [
        "# Plotar Boxplot\n",
        "fig , ax = plt.subplots(figsize=(8, 12), sharex=True)\n",
        "sns.boxplot(df.Class, df.Amount, showmeans=True, ax=ax, orient='v')\n",
        "\n",
        "# Definir limites\n",
        "plt.ylim((-10, 750))\n",
        "plt.xticks([0, 1],['Normal', 'Fraude'])\n",
        "\n",
        "# Mostrar Boxplot\n",
        "plt.tight_layout();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x864 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Pc3bJX9hnYkT"
      },
      "source": [
        "# **Apresentação dos Modelos**\n",
        "\n",
        "Para mostrar a diferença de desempenhho entre modelos treinados com dados balanceados e desbalanceados, irei construir 8 modelos:\n",
        "\n",
        "* 4 de **Regressão Logística**\n",
        "  * 1 com os dados desbalanceados;\n",
        "  * 3 com diferentes métodos de balanceamento de dados.\n",
        "* 4 de **Árvores de Decisão**\n",
        "  * 1 com os dados desbalanceados;\n",
        "  * 3 com diferentes métodos de balanceamento de dados.\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ULe7z0jZt0EH"
      },
      "source": [
        "## **Preparação dos Dados**\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JeuzbqQApojE"
      },
      "source": [
        "**Padronizar `Time` e `Amount`**\n",
        "\n",
        "Nessa etapa, iremos normalizar os dados da coluna `Time` e `Amount`, que possuíam seus valores originais. Para isso, a padronização será feita por meio da classe `StandardScaler`."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b6PuC4Pc-Pxv"
      },
      "source": [
        "# Importar pacotes nescessários do Scikit-Learn para preparar os dados\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import StandardScaler"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 224
        },
        "id": "UidxIZyU_fkA",
        "outputId": "0c8ee519-06e1-4cc6-fb77-613555044dac"
      },
      "source": [
        "# Criar cópia do DataFrame \n",
        "df_clean = df.copy()\n",
        "\n",
        "# Padronizar as colunas Time e Amount\n",
        "scaler = StandardScaler()\n",
        "df_clean['Scaler_amount'] = scaler.fit_transform(df_clean['Amount'].values.reshape(-1, 1))\n",
        "df_clean['Scaler_time'] = scaler.fit_transform(df_clean['Time'].values.reshape(-1, 1))\n",
        "\n",
        "df_clean.drop(['Time', 'Amount'], axis=1, inplace=True)\n",
        "\n",
        "# Ver as primeiras entradas\n",
        "df_clean.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>V1</th>\n",
              "      <th>V2</th>\n",
              "      <th>V3</th>\n",
              "      <th>V4</th>\n",
              "      <th>V5</th>\n",
              "      <th>V6</th>\n",
              "      <th>V7</th>\n",
              "      <th>V8</th>\n",
              "      <th>V9</th>\n",
              "      <th>V10</th>\n",
              "      <th>V11</th>\n",
              "      <th>V12</th>\n",
              "      <th>V13</th>\n",
              "      <th>V14</th>\n",
              "      <th>V15</th>\n",
              "      <th>V16</th>\n",
              "      <th>V17</th>\n",
              "      <th>V18</th>\n",
              "      <th>V19</th>\n",
              "      <th>V20</th>\n",
              "      <th>V21</th>\n",
              "      <th>V22</th>\n",
              "      <th>V23</th>\n",
              "      <th>V24</th>\n",
              "      <th>V25</th>\n",
              "      <th>V26</th>\n",
              "      <th>V27</th>\n",
              "      <th>V28</th>\n",
              "      <th>Class</th>\n",
              "      <th>Scaler_amount</th>\n",
              "      <th>Scaler_time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>-1.359807</td>\n",
              "      <td>-0.072781</td>\n",
              "      <td>2.536347</td>\n",
              "      <td>1.378155</td>\n",
              "      <td>-0.338321</td>\n",
              "      <td>0.462388</td>\n",
              "      <td>0.239599</td>\n",
              "      <td>0.098698</td>\n",
              "      <td>0.363787</td>\n",
              "      <td>0.090794</td>\n",
              "      <td>-0.551600</td>\n",
              "      <td>-0.617801</td>\n",
              "      <td>-0.991390</td>\n",
              "      <td>-0.311169</td>\n",
              "      <td>1.468177</td>\n",
              "      <td>-0.470401</td>\n",
              "      <td>0.207971</td>\n",
              "      <td>0.025791</td>\n",
              "      <td>0.403993</td>\n",
              "      <td>0.251412</td>\n",
              "      <td>-0.018307</td>\n",
              "      <td>0.277838</td>\n",
              "      <td>-0.110474</td>\n",
              "      <td>0.066928</td>\n",
              "      <td>0.128539</td>\n",
              "      <td>-0.189115</td>\n",
              "      <td>0.133558</td>\n",
              "      <td>-0.021053</td>\n",
              "      <td>0</td>\n",
              "      <td>0.244964</td>\n",
              "      <td>-1.996583</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1.191857</td>\n",
              "      <td>0.266151</td>\n",
              "      <td>0.166480</td>\n",
              "      <td>0.448154</td>\n",
              "      <td>0.060018</td>\n",
              "      <td>-0.082361</td>\n",
              "      <td>-0.078803</td>\n",
              "      <td>0.085102</td>\n",
              "      <td>-0.255425</td>\n",
              "      <td>-0.166974</td>\n",
              "      <td>1.612727</td>\n",
              "      <td>1.065235</td>\n",
              "      <td>0.489095</td>\n",
              "      <td>-0.143772</td>\n",
              "      <td>0.635558</td>\n",
              "      <td>0.463917</td>\n",
              "      <td>-0.114805</td>\n",
              "      <td>-0.183361</td>\n",
              "      <td>-0.145783</td>\n",
              "      <td>-0.069083</td>\n",
              "      <td>-0.225775</td>\n",
              "      <td>-0.638672</td>\n",
              "      <td>0.101288</td>\n",
              "      <td>-0.339846</td>\n",
              "      <td>0.167170</td>\n",
              "      <td>0.125895</td>\n",
              "      <td>-0.008983</td>\n",
              "      <td>0.014724</td>\n",
              "      <td>0</td>\n",
              "      <td>-0.342475</td>\n",
              "      <td>-1.996583</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>-1.358354</td>\n",
              "      <td>-1.340163</td>\n",
              "      <td>1.773209</td>\n",
              "      <td>0.379780</td>\n",
              "      <td>-0.503198</td>\n",
              "      <td>1.800499</td>\n",
              "      <td>0.791461</td>\n",
              "      <td>0.247676</td>\n",
              "      <td>-1.514654</td>\n",
              "      <td>0.207643</td>\n",
              "      <td>0.624501</td>\n",
              "      <td>0.066084</td>\n",
              "      <td>0.717293</td>\n",
              "      <td>-0.165946</td>\n",
              "      <td>2.345865</td>\n",
              "      <td>-2.890083</td>\n",
              "      <td>1.109969</td>\n",
              "      <td>-0.121359</td>\n",
              "      <td>-2.261857</td>\n",
              "      <td>0.524980</td>\n",
              "      <td>0.247998</td>\n",
              "      <td>0.771679</td>\n",
              "      <td>0.909412</td>\n",
              "      <td>-0.689281</td>\n",
              "      <td>-0.327642</td>\n",
              "      <td>-0.139097</td>\n",
              "      <td>-0.055353</td>\n",
              "      <td>-0.059752</td>\n",
              "      <td>0</td>\n",
              "      <td>1.160686</td>\n",
              "      <td>-1.996562</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>-0.966272</td>\n",
              "      <td>-0.185226</td>\n",
              "      <td>1.792993</td>\n",
              "      <td>-0.863291</td>\n",
              "      <td>-0.010309</td>\n",
              "      <td>1.247203</td>\n",
              "      <td>0.237609</td>\n",
              "      <td>0.377436</td>\n",
              "      <td>-1.387024</td>\n",
              "      <td>-0.054952</td>\n",
              "      <td>-0.226487</td>\n",
              "      <td>0.178228</td>\n",
              "      <td>0.507757</td>\n",
              "      <td>-0.287924</td>\n",
              "      <td>-0.631418</td>\n",
              "      <td>-1.059647</td>\n",
              "      <td>-0.684093</td>\n",
              "      <td>1.965775</td>\n",
              "      <td>-1.232622</td>\n",
              "      <td>-0.208038</td>\n",
              "      <td>-0.108300</td>\n",
              "      <td>0.005274</td>\n",
              "      <td>-0.190321</td>\n",
              "      <td>-1.175575</td>\n",
              "      <td>0.647376</td>\n",
              "      <td>-0.221929</td>\n",
              "      <td>0.062723</td>\n",
              "      <td>0.061458</td>\n",
              "      <td>0</td>\n",
              "      <td>0.140534</td>\n",
              "      <td>-1.996562</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>-1.158233</td>\n",
              "      <td>0.877737</td>\n",
              "      <td>1.548718</td>\n",
              "      <td>0.403034</td>\n",
              "      <td>-0.407193</td>\n",
              "      <td>0.095921</td>\n",
              "      <td>0.592941</td>\n",
              "      <td>-0.270533</td>\n",
              "      <td>0.817739</td>\n",
              "      <td>0.753074</td>\n",
              "      <td>-0.822843</td>\n",
              "      <td>0.538196</td>\n",
              "      <td>1.345852</td>\n",
              "      <td>-1.119670</td>\n",
              "      <td>0.175121</td>\n",
              "      <td>-0.451449</td>\n",
              "      <td>-0.237033</td>\n",
              "      <td>-0.038195</td>\n",
              "      <td>0.803487</td>\n",
              "      <td>0.408542</td>\n",
              "      <td>-0.009431</td>\n",
              "      <td>0.798278</td>\n",
              "      <td>-0.137458</td>\n",
              "      <td>0.141267</td>\n",
              "      <td>-0.206010</td>\n",
              "      <td>0.502292</td>\n",
              "      <td>0.219422</td>\n",
              "      <td>0.215153</td>\n",
              "      <td>0</td>\n",
              "      <td>-0.073403</td>\n",
              "      <td>-1.996541</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "         V1        V2        V3  ...  Class  Scaler_amount  Scaler_time\n",
              "0 -1.359807 -0.072781  2.536347  ...      0       0.244964    -1.996583\n",
              "1  1.191857  0.266151  0.166480  ...      0      -0.342475    -1.996583\n",
              "2 -1.358354 -1.340163  1.773209  ...      0       1.160686    -1.996562\n",
              "3 -0.966272 -0.185226  1.792993  ...      0       0.140534    -1.996562\n",
              "4 -1.158233  0.877737  1.548718  ...      0      -0.073403    -1.996541\n",
              "\n",
              "[5 rows x 31 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "St8yHqCUpw3A"
      },
      "source": [
        "**Separar dados entre Treino e Teste**\n",
        "\n",
        "Iremos também dividir o conjunto de dados entre treino e teste, assim poderemos testar o modelo com uma base de dados que ele não teve contato."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "D8CNC7JUDSis"
      },
      "source": [
        "# Separar os dados entre feature matrix e target vector \n",
        "X = df_clean.drop('Class', axis=1)\n",
        "y = df_clean['Class']\n",
        "\n",
        "# Dividir o dataset entre treino e teste\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, stratify=y, shuffle=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iPMvlZ4lnZpv"
      },
      "source": [
        "**Balanceamento dos dados**\n",
        "\n",
        "E por último iremos balancear os dados, de forma que nossos modelos apresentem um melhor desempenho na hora de identificar transações fraudulentas. Dessa forma, também evitamos o *overfitting* (quando um modelo se torna muito bom em dados ja vistos, mas não tem um bom desempenho em dados novos)\n",
        "\n",
        "Serão utilizados 3 métodos de balanceamento:\n",
        "\n",
        "* **RandomUnderSampling (RUS)** - Este método descarta um subjconjunto aleatório da classe majoritária, preservando as características da classe minoritária, ideal para quando se há um grande volume de dados. Este método pode acarretar em uma performance inferior na hora de fazer predições da classe majoritária.\n",
        "\n",
        "* **ADASYN** - Gera novas amostras, próximas às originais que são classificadas de forma errada usando um classificador K-Nearest Neighbors. Para entender melhor este método, [clique aqui](https://imbalanced-learn.org/stable/over_sampling.html#from-random-over-sampling-to-smote-and-adasyn). \n",
        "\n",
        "* **SMOTE** - Enquanto isso a implementação básica do **SMOTE**, não fará distinção entre amostras fáceis e difíceis de serem classificadas. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Tdw9oVJX-Jf0"
      },
      "source": [
        "### Balanceamento dos Dados com Under-Sampling (RUS)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 316
        },
        "id": "s9PYUgkzDfnY",
        "outputId": "d86463cc-969a-4b2d-ce95-f437d34dd333"
      },
      "source": [
        "# Balanceamento dos dados\n",
        "from imblearn.under_sampling import RandomUnderSampler\n",
        "rus = RandomUnderSampler()\n",
        "X_rus, y_rus = rus.fit_sample(X_train, y_train)\n",
        "\n",
        "# Checar o balanceamento das classes\n",
        "print(pd.Series(y_rus).value_counts())\n",
        "\n",
        "# Plotar a nova distribuição de Classes\n",
        "sns.countplot(y_rus);"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1    369\n",
            "0    369\n",
            "dtype: int64\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "U5CFgZ2y-fK7"
      },
      "source": [
        "### Balanceamento dos Dados com Over-Sampling (SMOTE) \n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 316
        },
        "id": "ncc5Ie3g-5gc",
        "outputId": "6af3bf04-7dd6-4585-c1c1-af67bc0ec339"
      },
      "source": [
        "# Balanceamento dos dados\n",
        "from imblearn.over_sampling import SMOTE\n",
        "smo = SMOTE()\n",
        "X_smo, y_smo = smo.fit_resample(X_train, y_train)\n",
        "\n",
        "# Checar o balanceamento das classes\n",
        "print(pd.Series(y_smo).value_counts())\n",
        "\n",
        "# Plotar a nova distribuição de Classes\n",
        "sns.countplot(y_smo);"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1    213236\n",
            "0    213236\n",
            "dtype: int64\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Lkz4raplCTm0"
      },
      "source": [
        "### Balanceamento dos Dados com Over-Sampling (ADASYN)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 316
        },
        "id": "xALLnqk8CYq-",
        "outputId": "3f6a12e5-5909-4d36-e5b2-1b6304a01343"
      },
      "source": [
        "# Balanceamento dos dados\n",
        "from imblearn.over_sampling import ADASYN\n",
        "ada = ADASYN()\n",
        "X_ada, y_ada = ada.fit_resample(X_train, y_train)\n",
        "\n",
        "# Checar o balanceamento das classes\n",
        "print(pd.Series(y_ada).value_counts())\n",
        "\n",
        "# Plotar a nova distribuição de Classes\n",
        "sns.countplot(y_ada);"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1    213243\n",
            "0    213236\n",
            "dtype: int64\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "thWBHxQs4yKp"
      },
      "source": [
        "Após balancear os dados podemos plotar matrizes de correlação e identificar quais variáveis estão mais relacionadas com as outras e verificar como cada balanceamento influenciou nessa correlação.\n",
        "\n",
        "Antes do balanceamento, não era possível obter informações relevantes da matriz de correlação, agora podemos verificar como cada variável se comporta em função de outra.\n",
        "\n",
        "Podemos perceber também que a relação não muda independentemente se os dados foram balanceados com RUS ou com SMOTE, mas quando olhamos para a matriz dos dados balanceados com ADASYN é possível notar uma diferença clara entre a relação das variáveis."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 347
        },
        "id": "L2SlZAfsWw_l",
        "outputId": "97f6ad39-b9e7-4514-c5f6-ecfab9a504f9"
      },
      "source": [
        "# plotar a matriz de correlação\n",
        "corr = X_train.corr()\n",
        "corr_rus = pd.DataFrame(X_rus).corr()\n",
        "corr_smo = pd.DataFrame(X_smo).corr()\n",
        "corr_ada = pd.DataFrame(X_ada).corr()\n",
        "\n",
        "fig, ax = plt.subplots(nrows=1, ncols=4, figsize = (30,8))\n",
        "fig.suptitle('Matriz de Correlação')\n",
        "\n",
        "sns.heatmap(corr, xticklabels=corr.columns, yticklabels=corr.columns,\n",
        "            linewidths=.1, cmap=\"coolwarm\", ax=ax[0])\n",
        "ax[0].set_title('Desbalanceado')\n",
        "\n",
        "sns.heatmap(corr_rus, xticklabels=corr.columns, yticklabels=corr.columns,\n",
        "            linewidths=.1, cmap=\"coolwarm\", ax=ax[1])\n",
        "ax[1].set_title('Balanceado com RUS')\n",
        "\n",
        "sns.heatmap(corr_smo, xticklabels=corr.columns, yticklabels=corr.columns,\n",
        "            linewidths=.1, cmap=\"coolwarm\", ax=ax[2])\n",
        "ax[2].set_title('Balanceado com SMOTE')\n",
        "\n",
        "sns.heatmap(corr_ada, xticklabels=corr.columns, yticklabels=corr.columns,\n",
        "            linewidths=.1, cmap=\"coolwarm\", ax=ax[3])\n",
        "ax[3].set_title('Balanceado com ADASYN')\n",
        "\n",
        "plt.tight_layout()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 2160x576 with 8 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "V74qWZAHxmQu"
      },
      "source": [
        "# **Machine Learning**\n",
        "\n",
        "Primeiro, vamos começar entendendo o que é *Machine Learning*.\n",
        "\n",
        "O algoritmo de aprendizagem de máquina (Machine Learning), também chamado de modelo, é uma expressão matemática que representa informação ou dados no contexto de qualquer problema em particular, que muitas vezes é um problema comercial. O principal objetivo é **ir dos dados para a percepção**, por meio do algoritmo.\n",
        "\n",
        "Antes de continuarmos, vamos deixar claro que existem duas categorias de aprendizagem de máquina: **supervisionada** e **não supervisionada**.\n",
        "\n",
        "Neste projeto, trabalharemos apenas com **modelos supervisionados**. \n",
        "\n",
        "Modelos supervisionados, são usados quando queremos **explicar ou prever dados**, isso é feito com a ajuda de dados antigos que serão destinados ao treino do modelo e assim ele será capaz de prever dados de saída para novas entradas."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Gg3BaUXScN4E"
      },
      "source": [
        "## **Regressão Logística**\n",
        "\n",
        "A Regressão Logística é um algoritmo de aprendizado de máquina que é usado para problemas de classificação, é um algoritmo de análise preditiva e se baseia no conceito de probabilidade.\n",
        "\n",
        "Caso queira um entendimento aprofundado sobre regressão logística, [clique aqui!](https://towardsdatascience.com/introduction-to-logistic-regression-66248243c148)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7W9BcP1FWTY_"
      },
      "source": [
        "### Construção do Modelo sem Balanceamento dos Dados\n",
        "\n",
        "Antes de começarmos a construir este modelo, vamos relembrar como estão distribuídas as classes, essa será a distribuição com a qual ele irá treinar. Então, não se assuste caso haja um desempenho inferior."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 381
        },
        "id": "4hB9g9XSZwsT",
        "outputId": "b7bd62ee-d137-4adf-ccb4-e30f06f7a25c"
      },
      "source": [
        "# Checar o balanceamento das classes\n",
        "print(df.Class.value_counts())\n",
        "print('\\nFraudes representam {:.4f}% do dataset.\\n'.format((df[df.Class == 1].shape[0] / df.shape[0]) * 100))\n",
        "\n",
        "# Plotar gráfico de barras para as Classes\n",
        "sns.countplot('Class', data=df);"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0    284315\n",
            "1       492\n",
            "Name: Class, dtype: int64\n",
            "\n",
            "Fraudes representam 0.1727% do dataset.\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kWSLx6X-rr7u"
      },
      "source": [
        "Para construir o modelo, vamos escolher e importar diretamente do Scikit-learn.\n",
        "\n",
        "O processo de treino deste modelo, será feito apenas com os dados divididos entre treino e teste, mas não com dados balanceados."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "B2qBRPrqWTI9"
      },
      "source": [
        "# Escolher e importa um modelo\n",
        "from sklearn.linear_model import LogisticRegression\n",
        "\n",
        "# Escolher e instanciar os Hyperparameters\n",
        "model_log = LogisticRegression()\n",
        "\n",
        "# Fit do Modelo (Treinar)\n",
        "model_log.fit(X_train, y_train)\n",
        "\n",
        "# Fazer previsões em cima de novos dados \n",
        "y_pred_log = model_log.predict(X_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6F2hJWi1XvWb"
      },
      "source": [
        "#### Avaliando o Desempenho do Modelo\n",
        "\n",
        "Aqui vamos imprimir um **relatório de classificação**, a **área sob a curva (AUC)** e uma **matriz de confusão**. Isso nos ajudará a avaliar o desempenho do nosso modelo, de forma que poderemos comparar com os modelos seguintes e dizer qual possui um melhor desempenho na detecção de fraudes.\n",
        "\n",
        "Uma ótima métrica que podemos utilizar para ver a taxa de acertos de fraude é a coluna *recall* do relatório de classificação\n",
        "\n",
        "O **AUC** é uma das melhores métricas para avaliação seu modelo, uma vez que quando as previsões estão 100% erradas seu valor é zero e quando estão 100% certas seu valor é 1. Cada modelo vai apresentar um valor de AUC, isso nos auxilia a determinar o melhor modelo. Para entender melhor o AUC, [clique aqui!](https://medium.com/bio-data-blog/entenda-o-que-%C3%A9-auc-e-roc-nos-modelos-de-machine-learning-8191fb4df772)\n",
        "\n",
        "Abaixo podemos ver que:\n",
        "\n",
        "* Nosso modelo possui uma grande quantidade de falsos negativos (**46 transações fraudulentas** classificadas como comuns), o que não é nada bom, pois o cliente ou o banco terá de arcar com os custos.\n",
        "\n",
        "* Houve um ótimo desempenho para prever as transações comuns, mas um desempenho abaixo do esperado na hora de prever fraudes (**apenas 77**).\n",
        "\n",
        "* Em casos como este, a precisão e a acurácia não são nossa prioridade, uma vez que se o modelo errar mais para falsos positivos e menos para falsos negativos, já é um grande avanço. \n",
        "\n",
        "* O **AUC** deste modelo não foi dos piores, mas é possível nós melhorarmos esse **0.8130**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4qMRkB_jJH3w"
      },
      "source": [
        "# Instalar pacote para plotar matriz de confusão\n",
        "!pip install -q scikit-plot"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 499
        },
        "id": "C6DX7hvNX2JR",
        "outputId": "d360d995-5886-41a3-bba3-9eab889b34ae"
      },
      "source": [
        "# Importar pacotes para avaliar o desempenho do modelo \n",
        "from sklearn.metrics import classification_report\n",
        "import scikitplot as skplt\n",
        "\n",
        "# Plotar matriz de confusão \n",
        "skplt.metrics.plot_confusion_matrix(y_test, y_pred_log)\n",
        "\n",
        "# Imprimir Relatório de Classificação\n",
        "print('Relatório de Classificação:\\n', classification_report(y_test, y_pred_log, digits=4))\n",
        "\n",
        "# Imprimir área sob curva\n",
        "print('AUC: {:.4f}\\n'.format(roc_auc_score(y_test, y_pred_log)))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Relatório de Classificação:\n",
            "               precision    recall  f1-score   support\n",
            "\n",
            "           0     0.9994    0.9999    0.9996     71079\n",
            "           1     0.9059    0.6260    0.7404       123\n",
            "\n",
            "    accuracy                         0.9992     71202\n",
            "   macro avg     0.9526    0.8130    0.8700     71202\n",
            "weighted avg     0.9992    0.9992    0.9992     71202\n",
            "\n",
            "AUC: 0.8130\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUUAAAEWCAYAAADxboUEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deZhU1Z3/8Xc1i5goiNgQBBWM+FXiDhE0JHFFNCagUUSNoDE6GfeJTjSJvzBxmdE8v4niRiSIghoBjUaiSIuocYkoi0sU/UYUkEaUBlpUMJCGmj/uabi0XdV1oYqqrv688tyHe889deoUT/h6zr1nSaXTaUREJFJR7AqIiJQSBUURkRgFRRGRGAVFEZEYBUURkRgFRRGRGAXFMmNm25vZX8xslZk9uBXlnGlmT+azbsVgZk+Y2Yhi10Oaj5TGKRaHmZ0B/AzYB/gMeA243t1f2MpyzwIuBg5397qtrmiemdkRwDPAn939pFj6gUR/B3919yNyKOe/gL3c/UeFqam0VGopFoGZ/Qy4GfhvoAuwO3AHMDgPxe8B/KMUA2JMDXCYmXWKpY0A/pGvLzCzlJnp/9+SmFqK25iZdQCWAOe4e6PdWzPbDrgRGBqSJgNXuvva0NK6D7gJuBJYD/zS3e82s98AvwBSwFrgUmA3Yi0qM+sBLADauHudmZ0N/BqoBJYDV7v7/SH9J+4+IHzucGAUsDdR8LrU3f8W7j0LPA8cBRwAvASc4e7LG/lt9fV/DPi7u99uZq2AD4A7gaPqW4pmNgo4GegAvAtc5u7Pm9kgYErsd77n7geGerwIHAEcAuwPjAXuc/exZjYa6OzuPwzl3wj0BY5xd/1DEEAtxWI4DGgHPJIlz6+A/sBBwIHAocDVsftfIwoU3YBzgdvNrKO7jyRqfU5y9x3c/a5sFTGzrwK3AMe7+47A4URd2Ib5dgYeD3k7Ab8DHm/Q0jsDOAfoDLQFrsj23cAEYHg4Pw54E/iwQZ5ZRH8HOwN/BB40s3buPq3B7zww9pmzgPOBHYFFDcq7HNjfzM42s28T/d2NUECUOAXFba8TsLyJ7u2ZwDXuvszda4DfEP1jr/evcP9f7j4V+BywLazPBmA/M9ve3Ze6+1uN5Pke8K673+vude7+APAO8P1Ynrvd/R/u/gVRy/agbF8aWpk7m5kRBccJjeS5z91XhO/8X2A7mv6d97j7W+Ez/2pQ3hqiv8ffEbVWL3b36ibKkxZGQXHbWwHsYmats+TZlc1bOYtC2sYyGgTVNcAOSSvi7quB04CfAkvN7HEz2yeH+tTXqVvs+qMtqM+9wEXAkTTScjazK8zs7fAm/ROi1vEuTZS5ONtNd38ZeJ+o6z05hzpKC6OguO29RPQcbEiWPB8SvTCptztf7lrmajXwldj11+I33b3K3Y8FuhK1/v6QQ33q67RkC+tU717gAmBqaMVtFLq3Pyd6rtrR3XcCVhEFM4BMXd6sXWEzu5CoxflhKF9kM9laK1IA7r7KzH5N9BywDniSqDt8DHCku/8ceAC42sxmEf0j/zVRd29LvAZcaWa7EwWVX9TfMLMuRM8unwK+IOqGb2ikjKnArWEY0WTgh0BvopclW8zdF5jZd4labg3tCNQRvalubWZXAe1j9z8GjjWzCndvrM5fYmZ7A9cRvYhZA7xiZk+4+5eeo0rLpZZiEYTnYz8jenlSQ9Tluwj4c8hyHTAbeAP4OzA3pG3Jd00HJoWy5rB5IKsI9fgQWAl8F/j3RspYAZxI9KJiBVEL68TG3i5vQf1ecPfGWsFVwDSiN92LgH+yede4/s39CjOb29T3hMcV9wE3uvvr7v4u8Evg3vC2XwTQkBwRkc2opSgiEqOgKCISo6AoIhKjoCgiElNSQ3KW136eXrR0ZbGrIQkcvO/uxa6CJJACUqmNYz23yJMvzkt36pjbXIE+vXevAgY1di/MZpoUS9qTaPjZhJDeA1gIDHX3WjNLEc2/P4FoSNXZ7j43lDWCTVNhr3P38SG9D3APsD3R0LJLm5rWWVJBcdHSlQw487fFroYkUDvrtmJXQRJo24qti4hAp4475Pzv9ItXb8s4A8ndnTAdNCwKsoRoZtNVwAx3vyGMT72KaPGT44Fe4egHjAb6hbn5I4kW90gDc8xsirvXhjznAS8TBcVBwBPZ6qzus4gkl6rI7cjd0USrHS0iWkJvfEgfz6bZX4OBCe6edveZwE5m1pVoQZHp7r4yBMLpwKBwr727zwytwwlkn0kGlFhLUUSagxRUtMp3ocOIZnIBdHH3peH8I6I1RyGaax8fwF8d0rKlVzeSnpWCoogkl8qtE15TU7PLgAEDZseSxrj7mHgeM2sL/IDYFNR67p42s206w0RBUUSSSZFz17iysnK5u/dtItvxwFx3/zhcf2xmXd19aegCLwvpS4gWTa7XPaQtIZrPHk9/NqR3byR/VnqmKCLJpVK5Hbk5nU1dZ4hWVa/fbGwE8GgsfXjYaqI/sCp0s6uAgWbW0cw6AgOBqnDvUzPrH95cD4+VlZGCoogklMrbi5aw+vuxwMOx5BuIVkB6l2j1qBtC+lSiFZXmEy1xdwGAu68EriVaqX0W0QLM9WP7LiDakmI+8B5NvHkGdZ9FZEvk3grMKix03KlB2gqit9EN86aBCzOUMw4Y10j6bGC/JHVSUBSRZFIFeftcMhQURSS5ZGMQmxUFRRFJLk/d51KkoCgiCaXUUhQR2YyCoohIkEpBK71oERHZRM8URURi1H0WEQnCSrXlSkFRRBLS22cRkc2ppSgiUk/T/ERENkmwnmJzpKAoIsmp+ywiUk8vWkRENqegKCISaD1FEZEG9ExRRCRG3WcRkRi1FEVEIilSpBQURUSCFAqKIiJxqYr8BEUz24loX+b9gDTwY8CBSUAPYCEw1N1rw4b2o4ATgDXA2e4+N5QzArg6FHudu48P6X2Ae4DtifaNvjRslZpR+T4tFZGCSaVSOR05GAVMc/d9gAOBt4GrgBnu3guYEa4Bjgd6heN8YDSAme0MjAT6AYcCI82sY/jMaOC82OcGNVUhBUURSSwfQdHMOgDfAe4CcPd17v4JMBgYH7KNB4aE88HABHdPu/tMYCcz6wocB0x395XuXgtMBwaFe+3dfWZoHU6IlZWRus8ikkiCViA1NTW7DBgwYHYsaYy7jwnnPYEa4G4zOxCYA1wKdHH3pSHPR0CXcN4NWBwrqzqkZUuvbiQ9KwVFEUkux0eKlZWVy929b4bbrYFDgIvd/WUzG8WmrjIA7p42s6zPAPNN3WcRSSxPzxSrgWp3fzlcP0QUJD8OXV/Cn8vC/SXAbrHPdw9p2dK7N5KelYKiiCRWUVGR05GNu38ELDYzC0lHA/OAKcCIkDYCeDScTwGGm1nKzPoDq0I3uwoYaGYdwwuWgUBVuPepmfUPb66Hx8rKSN1nEUkkld9xihcD95tZW+B94ByixtpkMzsXWAQMDXmnEg3HmU80JOccAHdfaWbXArNCvmvcfWU4v4BNQ3KeCEdWCooiklyeYqK7vwY09szx6EbypoELM5QzDhjXSPpsojGQOVNQFJGENM1PRGQzCooiIvVS+ZvmV4oUFEUkMbUURUSCFAqKIiIxetEiIrKJ1lMUEWmgfGOigqKIJNfUFL7mTEFRRBLRHi0iInEp1H0W6LVHZ+698ccbr3t268S1ox/nw2Wr+NVPT2Cfnl349ln/n7nzPtiY54ofD+TswYexfsMGLv/tQzz10tsZy7ntj89y8jEHZyxLto1bbr6Je+4eSyqV4hv77c+YsXfTrl27Yler5KiluIXMbBDRHgytgLHufkMhv6+Q3l20jP7DoupXVKR4r+p6pjzzOtu3a8uwy//AbVefvln+ffb8GqcedwiHnHI9XSs7MPX3F7H/kGsylgPw1nsfNlqWbBtLlizhjttv4dU35rH99ttz5ulDeXDSRM4acXaxq1ZyFBS3gJm1Am4HjiVaTHKWmU1x93mF+s5t5chDjQXVNXywtDZjnhOPOIAHq+ay7l91LPpwBe8tXs439+vBy28syFiOL/i44HWX7Orq6vjiiy9o06YNX6xZQ9dddy12lUpSOQfFQr5COhSY7+7vu/s6YCLRxjPN3qnH9WHytDlZ83Sr7ED1R5uC5pJltezauUPicmTb6datG5f9xxXsvefu9NytK+3bd+CYYwcWu1qlJ8x9zuVojgoZFDNtJtOstWndiu99d38env5qSZQj+VNbW8tjf3mUt99dwPsffMjqNat54P77il2tklP/9jlPW5yWnPIdbFQgxw3ozWvvLGbZys+y5ltSs4ruX+u48bpb5458uGxV4nJk23l6xlP06NGTyspK2rRpw5AhJzPzpb8Vu1olSUFxy2TaTKZZGzqob05d3seffYNTjzuEtm1as8eundhr90pmvbkwcTmy7ey22+688spM1qxZQzqd5pmnZ2D77FvsapWkaEuCpo/mqJBBcRbQy8x6hv0XhhFtPNNsfaVdW47qtw+PPv3axrQfHHkA86ddS78DevDwLT9lyu3Raulvv/8Rf3ryVV7906+YcvsFXHbDZDZsSGcsJ1tZsm0c2q8fJ518Cocdegh9D96fDRs2cO555xe7WiWpnFuKqXS6cFuqmtkJwM1EQ3LGufv12fLPmfdBesCZvy1YfST/amfdVuwqSAJtW0FFauuGXr9ZvSr9w1tn5pTXbzxuDo3vwVKyCjpO0d2nEu3AJSJlpJk2AnOiGS0iklhFMx1ukwsFRRFJJo8vUcxsIfAZsB6oc/e+ZrYzMAnoASwEhrp7bdjQfhTR3s9rgLPdfW4oZwRwdSj2OncfH9L7sGnf56nApWGr1Iw0JEdEEinAOMUj3f0gd69/9ngVMMPdewEzwjXA8UCvcJwPjAYIQXQk0I9o0shIM6sfDzcaOC/2uUFNVUZBUUQSK/CQnMHA+HA+HhgSS5/g7ml3nwnsZGZdgeOA6e6+0t1rgenAoHCvvbvPDK3DCbGyMlL3WUQSy3WR2Zqaml0GDBgwO5Y0xt3HxK7TwJNmlgbuDPe6uPvScP8joEs4zzRLLlt6dSPpWSkoikgyCVqBlZWVy2Pd4sYMcPclZtYZmG5m78Rvuns6BMxtRt1nEUmkfovTfDxTdPcl4c9lwCNEzwQ/Dl1fwp/LQvZMs+SypXdvJD0rBUURSSwfzxTN7KtmtmP9OTAQeJNo5tuIkG0E8Gg4nwIMN7OUmfUHVoVudhUw0Mw6hhcsA4GqcO9TM+sf3lwPj5WVkbrPIpJYnqbwdQEeMTOIYtEf3X2amc0CJpvZucAiYGjIP5VoOM58oiE55wC4+0ozu5ZoajHANe6+MpxfwKYhOU+EIysFRRFJLB8x0d3fBw5sJH0FcHQj6Wmg0QUB3H0cMK6R9NnAfknqpaAoIomkUinNaBERiWuuK+DkQkFRRBIr45iooCgiyamlKCJSrxmvqp0LBUURSSSFlg4TEdmMus8iIhs13/1XcqGgKCLJ6JmiiMgm9QtClCsFRRFJrIxjooKiiCSnt88iIkEqBRVl3FRUUBSRxMo4JiooikhyetEiIhJTxo8UMwdFM7uVaKetRrn7JQWpkYiUtJa8nuLsLPdEpAVL0QKDoruPj1+b2VfcfU3hqyQipa6MG4pN7+ZnZoeZ2TzgnXB9oJndUfCaiUhJyucWp6Uoly1ObwaOA1YAuPvrwHcKWSkRKW352OK0VOX09tndF4dtCOutL0x1RKQ5yOfgbTNrRfQOY4m7n2hmPYGJQCdgDnCWu68zs+2ACUAfokbaae6+MJTxC+Bcoth0ibtXhfRBwCigFTDW3W9o8rflUOfFZnY4kDazNmZ2BfB2kh8tIuUjlYqm+eVy5OhSNo8pNwI3ufteQC1RsCP8WRvSbwr5MLPewDDgG8Ag4A4zaxWC7e3A8UBv4PSQN6tcguJPifZa7QZ8CBxEhr1XRaRlyFf32cy6A98DxobrFHAU8FDIMh4YEs4Hh2vC/aND/sHARHdf6+4LgPnAoeGY7+7vu/s6otbn4Kbq1GT32d2XA2c2/fNEpGVI5dx9rqmp2WXAgAHx4X1j3H1M7Ppm4OfAjuG6E/CJu9eF62qiBhnhz8UA7l5nZqtC/m7AzFiZ8c8sbpDer6k6NxkUzWxPoj55f6LB3C8B/+Hu7zf1WREpT7l2jCsrK5e7e9/G7pnZicAyd59jZkfkrXJbKZfu8x+ByUBXYFfgQeCBQlZKREpXHofkfAv4gZktJOraHkXUANvJzOobbN2BJeF8CbAbQLjfgeiFy8b0Bp/JlJ5VLkHxK+5+r7vXheM+oF0OnxORcpSKBm/ncmTj7r9w9+7u3oPoRcnT7n4m8AxwSsg2Ang0nE8J14T7T7t7OqQPM7PtwpvrXsArwCygl5n1NLO24TumNPXzss193jmcPmFmVxFF8jRwGjC1qYJFpHwVeO7zlcBEM7sOeBW4K6TfBdxrZvOBlURBDnd/y8wmA/OAOuBCd18PYGYXAVVEQ3LGuftbTX15Kp1ufM0HM1tAFAQb+/Vpd98z55+YoznzPkgPOPO3+S5WCqh21m3FroIk0LYVVGzlxOUFK9akR1bNzynvhDMOmAM0+kyxVGWb+9xzW1ZERJqPcp77nNOMFjPbj2jw48Znie4+oVCVEpESlmrhi8ya2UjgCKKgOJVodPgLRNNtRKSFSZH7kJzmKJe3z6cARwMfufs5wIFEr8JFpEVK0aoit6M5yiUofuHuG4A6M2sPLGPzsT8i0sKU89JhuTxTnG1mOwF/IFqx4nOiWS0i0gJFg7eLXYvCyWXu8wXh9PdmNg1o7+5vFLZaIlKyWuq+z2Z2SLZ77j63MFUSkVJXxjExa0vxf7PcSxPNU8yrg/fdXYOBRZqB5vq8MBfZBm8fuS0rIiLNQwpo1RKDoohIo3JY7KE5U1AUkURSKCiKiMQ03zGIuchlml+KaDuCPd39GjPbHfiau79S8NqJSEkq55ZiLjNa7gAOA04P158R7ZAlIi1Q/eDtct33OZeg2M/dLwT+CeDutUDbgtZKREpXClqnUjkdzVEuzxT/FfZPTQOYWSWwoaC1EpGS1eKn+QG3AI8Anc3seqJVc64uaK1EpKS1yGl+9dz9fjObQ7R8WAoY4u5vF7xmIlKyyjgmNv1MMbxtXgP8hWgnrNUhTURaoPpxilu7m1+pyqX7/DibNrBqB/QEHPhGAeslIqUqRbNdQDYXuXSf949fh9VzLsiQXUTKnGa0NODuc82sXyEqIyLNw1bukgqAmbUDngO2I4pFD7n7yLCh/USgE9HC1me5+zoz245ob6g+wArgNHdfGMr6BXAusB64xN2rQvogYBTRvs9j3f2GpuqVy4yWn8UuK4BDgA9z+dEiUp7y1FJcCxzl7p+bWRvgBTN7AvgZcJO7TzSz3xMFu9Hhz1p338vMhgE3AqeZWW9gGNEjvV2Bp8xs7/AdtwPHAtXALDOb4u7zsv62HCq+Y+zYjugZ4+Akv1xEyke+XrS4e9rdPw+XbcJRv1brQyF9PDAknA8O14T7R4dpyIOBie6+1t0XAPOBQ8Mx393fd/d1RK3PJmNX1pZiGLS9o7tf0VRBItJCJNiUqqamZpcBAwbMjiWNcfcx9RchxswB9iJq1b0HfOLudSFLNdAtnHcDFgO4e52ZrSLqYncDZsa+I/6ZxQ3Sm3z0l207gtbhi7/VVCEi0rK0yqWPCVRWVi53976Z7rv7euCgsDneI8A+eangVsjWUnyF6Pnha2Y2BXgQWF1/090fLnDdRKQERd3n/L5+dvdPzOwZosVndqpvlAHdgSUh2xKi7ZWrzaw10f7zK2Lp9eKfyZSeUS7xvl344qOAE4Hvhz9FpIXKxzNFM6sMLUTMbHuiFyJvA88QTScGGAE8Gs6nhGvC/afdPR3Sh5nZduHNdS+iRt0soJeZ9TSztkQvY6Y09duytRQ7hzfPb7Jp8Ha9dFMFi0iZyt+yYF2B8eG5YgUw2d0fM7N5wEQzuw54Fbgr5L8LuNfM5gMriYIc7v6WmU0G5gF1wIWhW46ZXQRUEQ3JGefubzX589LpxuObmS0leg3e2M9Pu/s1uf3u3G1Ik163Pt+liki9tq2gYisHGS77fG36wdeX5pT3wm/1mANkfKZYirK1FJcWIvCJSPNXzgtCZAuKZfyzRWRLpYDWZTzPL1tQPHqb1UJEmpUW2VJ095XbsiIi0kykWvgisyIicdqOQESkgRwntDRLCooiklBK3WcRkXqFmOZXShQURSSx8g2JCooisgXKuKGooCgiCaXIeT3F5khBUUQSSaG3zyIim9GLFhGRGHWfRUQCdZ9FRDaT+8ZVzZGCoogkVr4hUUFRRBJKAa3UUhQRCfK3R0tJUlAUkcS2cpuXkqagKCKJaD1FEZEGKvLQUjSz3YAJQBeibZPHuPsoM9sZmAT0ABYCQ9291sxSwCjgBGANcLa7zw1ljQCuDkVf5+7jQ3of4B5ge2AqcGnYKzrLbxMRSSiVyu1oQh1wubv3BvoDF5pZb+AqYIa79wJmhGuA44k2uu8FnE+0BTMhiI4E+gGHAiPNrGP4zGjgvNjnBjVVKQVFEUmkfj3FXI5s3H1pfUvP3T8D3ga6AYOB8SHbeGBIOB8MTHD3tLvPBHYys67AccB0d1/p7rXAdGBQuNfe3WeG1uGEWFkZqfssIonle4dTM+sBHAy8DHRx96Xh1kdE3WuIAubi2MeqQ1q29OpG0rNSS1FEEkrl/L+amppdzGx27Di/YWlmtgPwJ+Ayd/80fi+08LI+A8w3tRRFJJkE4xQrKyuXu3vfTPfNrA1RQLzf3R8OyR+bWVd3Xxq6wMtC+hJgt9jHu4e0JcARDdKfDendG8mflVqKBbB+/Xr69z2YkwefCEA6nWbk//sV+/fem4P235fbb72lyDWUev9wp1+fgzYenXduz62jbuZHZ5y2Mc326kG/PgcVu6olI0XubcVswtvku4C33f13sVtTgBHhfATwaCx9uJmlzKw/sCp0s6uAgWbWMbxgGQhUhXufmln/8F3DY2VlVLCWopmNA04Elrn7foX6nlJ02y2jsH335bNPo57AvePvoXrxYl5/8x0qKipYtmxZEyXItrK3GS/PeQ2I/mP29T268YMhJ3HxpZdtzHPlf15Ohw4dilXFkpSnZ4rfAs4C/m5mr4W0XwI3AJPN7FxgETA03JtKNBxnPtGQnHMA3H2lmV0LzAr5rnH3leH8AjYNyXkiHFkVsvt8D3Ab0RufFqO6upppTzzOlb/4FbfcHP3Hb8ydoxl/7x+pqIga5p07dy5mFSWDZ56eQc89v84ee+yxMS2dTvOnhyYz7cmni1iz0pOPRWbd/QUyry1xdCP508CFGcoaB4xrJH02kKhRVrDus7s/B6xsMmOZ+c/LL+P6//ntxgAIsOD993jowUl8q19fBp94PPPffbeINZRMHpw0kaGnnb5Z2osvPE+Xzl3Yq1evItWq9KQSHM2Rninm0dTHH6NzZWcO6dNns/S1a9eyXbt2vPjybM459zz+7bwfF6mGksm6det4/LEpnHzKqZulT574AKcOOz3Dp1qufIxTLFV6+5xHL/3tRR57bArTpk1l7T//yaeffso5w39Et+7dGTLkZAAGDzmJf/vJOUWuqTRUNe0JDjr4ELp06bIxra6ujkf//DAvvjyniDUrTc0z3OVGLcU8uvb6/+G9hdX4/IVMuH8iRxx5FHdPuI/v/2AIf332GQCef+6v7NVr7yLXVBqaPOmBL3Wdn57xFHvbPnTv3j3Dp1qwMu4/q6W4DVzx86s4Z/iZ3DrqJr66ww6MvnNssaskMatXr+bpp6Zz2x13bpbe2DNGicYoNteucS5S6XRhBoub2QNEAyp3AT4GRrr7Xdk+syFNet36glRHRIC2raBiKxdDXL22Lj1vyeqc8n5zzw5zgIyDt0tRwVqK7q7/xIqUq/JtKKr7LCJJNT1bpTlTUBSRxMr4kaKCoogkV8YxUUFRRJKJ9mgp37CooCgiiZVxTFRQFJHkyjgmKiiKSELNeLZKLhQURSQxDckREYnRM0URkSB6+1zsWhSOgqKIJKbus4hIjFqKIiIxZRwTFRRFZAuUcVRUUBSRRMp9kVkFRRFJLF8hsbH94c1sZ2AS0ANYCAx199qwof0oor2f1wBnu/vc8JkRwNWh2OvcfXxI78OmfZ+nApeGrVIz0h4tIpJc/vZouQcY1CDtKmCGu/cCZoRrgOOBXuE4HxgNG4PoSKAfcCgw0sw6hs+MBs6Lfa7hd32JgqKIJJTK+X9NybA//GBgfDgfDwyJpU9w97S7zwR2MrOuwHHAdHdf6e61wHRgULjX3t1nhtbhhFhZGan7LCKJ5fpIsaamZpcBAwbMjiWNcfcxTXysi7svDecfAfX7znYDFsfyVYe0bOnVjaRnpaAoIonl+kyxsrJyubtv8cZV7p42s8LsrpeBus8ikkj9IrO5HFvo49D1Jfy5LKQvAXaL5ese0rKld28kPSsFRRFJLJXK7dhCU4AR4XwE8GgsfbiZpcysP7AqdLOrgIFm1jG8YBkIVIV7n5pZ//DmenisrIzUfRaRxPI4JGfj/vBmVk30FvkGYLKZnQssAoaG7FOJhuPMJxqScw6Au680s2uBWSHfNe5e//LmAjYNyXkiHFml0ult2l3PakOa9Lr1xa6FSPlq2woqtnI1h7V1G9JLV63LKW+PTu3mAFv8TLEY1FIUkcS0So6ISEwZz/JTUBSRZFJAhYKiiEhc+UZFBUURSUzdZxGRmDKOiQqKIpKcWooiIkE0W6V8o6KCoogkVr4hUUFRRLZAGTcUFRRFJKncFpBtrhQURSS58o2JCooiklwZx0QFRRFJJprmV75hUUFRRBIr45iolbdFROLUUhSRZLZuq4GSp6AoIolpSI6ISBDt5lfsWhSOgqKIJKagKCISo+6ziEiMWooiIjFlHBMVFEVkC5RxVFRQFJFEUqmUdvPbVipSLG/XmkXFrodIGdtjawuoSFHVrjW75Jh9+dZ+37aWSqfTxa6DiEjJ0NxnEZEYBUURkRgFRRGRGAVFEZEYBUURkRgFRRGRmJIap1huzGwQMApoBYx19xuKXCVpgpmNA04Elrn7fsWuj2x7aikWiCNIlMkAAAO+SURBVJm1Am4Hjgd6A6ebWe/i1kpycA8wqNiVkOJRUCycQ4H57v6+u68DJgKDi1wnaYK7PwesLHY9pHgUFAunG7A4dl0d0kSkhCkoiojEKCgWzhJgt9h195AmIiVMb58LZxbQy8x6EgXDYcAZxa2SiDRFLcUCcfc64CKgCngbmOzubxW3VtIUM3sAeCk6tWozO7fYdZJtS0uHiYjEqKUoIhKjoCgiEqOgKCISo6AoIhKjoCgiEqOg2IyY2Xoze83M3jSzB83sK1tR1j1mdko4H5ttsQozO8LMDt+C71hoZl/a9S1TeoM8nyf8rv8ysyuS1lGkIQXF5uULdz8oLGm1Dvhp/KaZbdFgfHf/ibvPy5LlCCBxUBRpjjSjpfl6HjjAzI4ArgVqgX3MbF/gBqJAth1wu7vfaWYp4FbgWKKFKtbVF2RmzwJXuPvssAbkfxOtAbkcOJco+K43sx8BFwPvAL8Hdg9FXObuL5pZJ+ABooUvXgKa3DLdzP5MNB2yHTDK3cfE7t0EDAQ+Aoa5e42ZfZ1oSbZKYA1wnru/k+DvTSQrtRSbodAiPB74e0g6BLjU3fcmCmKr3P2bwDeB88JUw5MAI1rbcTiNtPzMrBL4A/BDdz8QONXdFxIFwJtCK/V5ooVzbwrf8UNgbChiJPCCu38DeIRNQTObH7t7H6AvcEkIrABfBWaHsv4aygYYA1wcPnMFcEcO3yGSM7UUm5ftzey1cP48cBdRcHvF3ReE9IFELchTwnUHoBfwHeABd18PfGhmTzdSfn/gufqy3D3TuoLHAL3NrP66vZntEL7j5PDZx82sNoffdImZnRTOdwt1XQFsACaF9PuAh8N3HA48GPvu7XL4DpGcKSg2L1+4+0HxhBAcVseSUkQtqaoG+U7IYz0qgP7u/s9G6pKz0PU/BjjM3deEbny7DNnT4Xs/afh3IJJP6j6Xnyrg382sDYCZ7W1mXwWeA04zs1Zm1hU4spHPzgS+E7rbmNnOIf0zYMdYvieJni0S8tUHqecIKwGZ2fFAxybq2gGoDQFxH6KWar0KoL61ewZRt/xTYIGZnRq+I2VmBzbxHSKJKCiWn7HAPGCumb0J3EnUI3gEeDfcm0D0ImQz7l4DnE/UVX2dTd3XvwAnheFA3wYuAfqa2RtmNo9Nb8F/QxRU3yLqRn/QRF2nAa3N7G2il0MzY/dWA4eG33AUcE1IPxM4N9TvLbTFg+SZVskREYlRS1FEJEZBUUQkRkFRRCRGQVFEJEZBUUQkRkFRRCRGQVFEJOb/AHPpAHtPjCuEAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wJNH5qcjRxfX"
      },
      "source": [
        "### Construção do Modelo (RUS)\n",
        "\n",
        "Aqui o processo é bem parecido, a grande diferença é que nosso modelo será treinado com dados balanceados, pelo método (RUS) e depois efetuará as previsões em cima dos dados de teste."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IqfjG_SUSTi-"
      },
      "source": [
        "# Escolher e instanciar os Hyperparameters\n",
        "model_log_rus = LogisticRegression()\n",
        "\n",
        "# Fit do modelo (Treinar)\n",
        "model_log_rus.fit(X_rus, y_rus)\n",
        "\n",
        "# Fazer previsões em cima de novos dados \n",
        "y_pred_rus = model_log_rus.predict(X_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3e4ENOTYSUXi"
      },
      "source": [
        "#### Avaliando o Desempenho do Modelo\n",
        "\n",
        "Dessa vez, nossas métricas nos dizem que:\n",
        "\n",
        "* A quantidade de falsos negativos diminuiu **consideravelmente** (de 46 para 10), o que é um ponto positivo. Dessa forma, o prejuízo do banco será menor.\n",
        "\n",
        "* Com a melhora na previsão de fraudes (**de 77 para 113**) quase o dobro, houve também um grande aumento nos falsos positivos (**de 8 para 2710**), mas isso já era esperado e não é o pior cenário possível.\n",
        "\n",
        "* Esse modelo teve um excelente **AUC** - **0.9403**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "H1jEi7gkSe2r",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 499
        },
        "outputId": "3d9d5de8-b041-46df-8a51-1ab8653c78d1"
      },
      "source": [
        "# Plotar matriz de confusão\n",
        "skplt.metrics.plot_confusion_matrix(y_test, y_pred_rus)\n",
        "\n",
        "# Imprimir relatório de classificação\n",
        "print('Relatório de Classificação:\\n',classification_report(y_test, y_pred_rus, digits=4))\n",
        "\n",
        "# Imprimir a área sob curva\n",
        "print('AUC: {:.4f}\\n'.format(roc_auc_score(y_test, y_pred_rus)))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Relatório de Classificação:\n",
            "               precision    recall  f1-score   support\n",
            "\n",
            "           0     0.9999    0.9619    0.9805     71079\n",
            "           1     0.0400    0.9187    0.0767       123\n",
            "\n",
            "    accuracy                         0.9618     71202\n",
            "   macro avg     0.5199    0.9403    0.5286     71202\n",
            "weighted avg     0.9982    0.9618    0.9789     71202\n",
            "\n",
            "AUC: 0.9403\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fz5EBlWzCKw-"
      },
      "source": [
        "### Construção do Modelo (SMOTE)\n",
        "\n",
        "Neste caso, nosso modelo será treinado com dados balanceados pelo método (SMOTE) e depois efetuará as previsões em cima dos dados de teste."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kXKBZLPVAXdR"
      },
      "source": [
        "# Escolher e importar um modelo\n",
        "from sklearn.linear_model import LogisticRegression\n",
        "\n",
        "# Escolher e instanciar os Hyperparameters\n",
        "model_log_smo = LogisticRegression()\n",
        "\n",
        "# Fit do modelo (Treinar)\n",
        "model_log_smo.fit(X_smo, y_smo)\n",
        "\n",
        "# Fazer previsões em cima de novos dados \n",
        "y_pred_smo = model_log_smo.predict(X_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PTiiVLfOAr4h"
      },
      "source": [
        "#### Avaliando o Desempenho Modelo\n",
        "\n",
        "Com este modelo, podemos ver que:\n",
        "\n",
        "* A quantidade de falsos negativos teve um aumento, quando comparado ao modelo (RUS), **foram 12 falsos negativos contra 10**.\n",
        "\n",
        "* A quantidade de fraudes previstas também sofreu alterações, **foram 111 contra 113 do modelo (RUS)**.\n",
        "\n",
        "* Porém, o diferencial deste modelo é que a quantidade de falsos positivos diminuiu bastante (**de 2710 para 1702**) uma diminuição de **mais de 1000 casos**, fazendo com que os clientes do banco que utilizasse esse modelo, tivessem seus cartões bloqueados, menos vezes.\n",
        "\n",
        "* Esse modelo teve um excelente **AUC** mas não o melhor : **0.9392**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 499
        },
        "id": "cc5SSIWWAp3n",
        "outputId": "e35d7410-2124-4db7-f4f8-607726b32841"
      },
      "source": [
        "# Plotar matriz de confusão\n",
        "skplt.metrics.plot_confusion_matrix(y_test, y_pred_smo)\n",
        "\n",
        "# Imprimir relatório de classificação\n",
        "print('Relatório de Classificação:\\n',classification_report(y_test, y_pred_smo, digits=4))\n",
        "\n",
        "# Imprimir a área sob curva\n",
        "print('AUC: {:.4f}\\n'.format(roc_auc_score(y_test, y_pred_smo)))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Relatório de Classificação:\n",
            "               precision    recall  f1-score   support\n",
            "\n",
            "           0     0.9998    0.9761    0.9878     71079\n",
            "           1     0.0612    0.9024    0.1147       123\n",
            "\n",
            "    accuracy                         0.9759     71202\n",
            "   macro avg     0.5305    0.9392    0.5512     71202\n",
            "weighted avg     0.9982    0.9759    0.9863     71202\n",
            "\n",
            "AUC: 0.9392\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "p-UgMZEqCr0i"
      },
      "source": [
        "### Construção do Modelo (ADASYN)\n",
        "\n",
        "Neste caso, nosso modelo será treinado com dados balanceados pelo método (ADASYN) e depois efetuará as previsões em cima dos dados de teste."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2NnjLRgGCvK4"
      },
      "source": [
        "# Escolher e plotar um modelo\n",
        "model_log_ada = LogisticRegression()\n",
        "\n",
        "# Fit do modelo (Treinar)\n",
        "model_log_ada.fit(X_ada, y_ada)\n",
        "\n",
        "# Fazer previsões em cima de novos dados \n",
        "y_pred_ada = model_log_ada.predict(X_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lqVjIm-pDPiG"
      },
      "source": [
        "#### Avaliando o Desempenho Modelo\n",
        "\n",
        "No nosso último modelo de regressão logística, os resultados apresentados são  interessantes, vamos dar uma olhada:\n",
        "\n",
        "* A quantidade de falsos negativos foi a menor dos nossos modelos de regressão logística, sendo apenas **6 falsos negativos**, contra as **10 do modelo (RUS)**.\n",
        "\n",
        "* A quantidade de fraudes previstas foi a maior entre todos os modelos, foram **117 fraudes previstas**, enquanto nos outros modelos não passava de **113 previsões**.\n",
        "\n",
        "* Porém, um ponto negativo deste modelo é que a quantidade de falsos positivos aumentou bastante, tendo a maior quantidade quando comparado aos outros modelos (sendo **6533 contra os 1702 do modelo SMOTE**), assim os clientes do banco que escolhesse esse modelo, teriam seus cartões bloqueados mais vezes, mas em compensação o prejuízo seria menor, caso seu cartão fosse fraudado.\n",
        "\n",
        "* Esse modelo teve o menor **AUC** dos modelos que treinaram com dados balanceados : **0.9297**\n",
        "\n",
        "Porém, vale lembrar que o AUC sozinho não é suficiente para dizer que esse é o melhor ou o pior modelo, pois depende de qual a solução mais eficiente para a instituição. Às vezes, a melhor solução para o banco é aquela que vai ter uma maior quantidade de falsos positivos e ter menos falsos negativos, do que o inverso."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 499
        },
        "id": "TyXZfkyiDTRu",
        "outputId": "085ed7f3-d4e1-48fd-dea1-97b899cc58be"
      },
      "source": [
        "# Plotar a matriz de confusão\n",
        "skplt.metrics.plot_confusion_matrix(y_test, y_pred_ada)\n",
        "\n",
        "# Imprimir relatório de classificação\n",
        "print('Relatório de Classificação: \\n', classification_report(y_test, y_pred_ada, digits=4))\n",
        "\n",
        "# Imprimir a área sob curva\n",
        "print('AUC: {:.4f}\\n'.format(roc_auc_score(y_test, y_pred_ada)))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Relatório de Classificação: \n",
            "               precision    recall  f1-score   support\n",
            "\n",
            "           0     0.9999    0.9081    0.9518     71079\n",
            "           1     0.0176    0.9512    0.0345       123\n",
            "\n",
            "    accuracy                         0.9082     71202\n",
            "   macro avg     0.5088    0.9297    0.4932     71202\n",
            "weighted avg     0.9982    0.9082    0.9502     71202\n",
            "\n",
            "AUC: 0.9297\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UbqAttBfcxjD"
      },
      "source": [
        "## **Árvores De Decisão (Decision Trees)**\n",
        "\n",
        "Os processos a seguir são extremamente parecidos com o que fizemos anteriormente, porém agora o modelo será uma árvore de decisão, sendo assim o método usado é diferente para classificar as transações.\n",
        "\n",
        "Dessa forma, seremos capazes de definir qual modelo seria melhor para Detecção de Fraude, uma árvore de decisão ou uma regressão logística.\n",
        "\n",
        "Para melhor compreensão, irei dar uma breve explicação de como um modelo de Árvore de Decisão funciona.\n",
        "\n",
        "As árvores de decisão são construídas a partir de um processo de indução, onde os dados são divididos em subconjuntos cada vez mais puros, dessa forma é encontrada a maior pureza em sua classificação. Para selecionar quais atributos serão utilizados na composição da árvore, é considerado o grau de entropia de um nó e a informação ganha após um split.\n",
        "\n",
        "Para uma visualização gráfica de como é construída uma árvore, acesse [o link do R2D3](http://www.r2d3.us/uma-introducao-visual-ao-aprendizado-de-maquina-1/)."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1mOeUV45hZjB"
      },
      "source": [
        "### Construção do Modelo sem Balanceamento dos Dados\n",
        "\n",
        "Para construir o modelo, mais uma vez vamos escolher e importar diretamente do Scikit-learn.\n",
        "\n",
        "O processo de treino deste modelo, será realizado apenas com os dados divididos entre treino e teste, mas não com dados balanceados."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VcsQzNTgdfP9"
      },
      "source": [
        "# Escolher e importar o modelo\n",
        "from sklearn.tree import DecisionTreeClassifier\n",
        "\n",
        "# Escolher os Hyperparameters\n",
        "model_tree = DecisionTreeClassifier(max_depth=4, criterion='entropy')\n",
        "\n",
        "# Fit do Modelo(Treinar)\n",
        "model_tree.fit(X_train, y_train)\n",
        "\n",
        "# Fazer previsões em cima dos dados de teste\n",
        "y_pred_tree = model_tree.predict(X_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "K0G5vdVOhPw_"
      },
      "source": [
        "#### Avaliando o Desempenho do Modelo\n",
        "\n",
        "Aqui vamos usar as mesmas métricas que você já está familiarizado, o **relatório de classificação**, a **área sob a curva (AUC)** e uma **matriz de confusão**. \n",
        "\n",
        "Lembre-se sempre que podemos ver a taxa de acertos de fraude na coluna *recall* do relatório de classificção.\n",
        "\n",
        "Vamos dar uma olhada em como nosso primeiro modelo de árvore de decisão se comportou:\n",
        "\n",
        "* Nosso modelo possui uma quantidade menor de falsos negativos que o primeiro modelo de RL (Regressão Logística) (**34 contra as 46 transações fraudulentas** classificadas como comuns), aqui podemos verificar que uma árvore de decisão teve um desempenho melhor com dados desbalanceados.\n",
        "\n",
        "* Houve um ótimo desempenho para prever as transações comuns, até mesmo melhor que o modelo de RL, mas um desempenho abaixo do que seria satisfatório (**89 contra 77**).\n",
        "\n",
        "* **Lembrando mais uma vez**, em casos como este que a precisão e a acurácia não são nossa prioridade, uma vez que se o modelo errar mais para falsos positivos e menos para falsos negativos, já é um grande avanço. \n",
        "\n",
        "* O **AUC** deste modelo foi melhor que o nosso modelo de RL : **0.8617**. \n",
        "\n",
        "Até o momento, tudo indica que modelos de **Árvore de Decisão são melhores** nessa situação, mas será que apenas com dados desbalanceados ou com balanceados também? \n",
        "\n",
        "Vamos checar isso logo abaixo."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 482
        },
        "id": "B4xVJAN7hqXE",
        "outputId": "64f7a74d-0a16-4b07-ef8c-f92e26d4783a"
      },
      "source": [
        "# Plotar matriz de confusão\n",
        "skplt.metrics.plot_confusion_matrix(y_test, y_pred_tree)\n",
        "\n",
        "# Imprimir classification report\n",
        "print(classification_report(y_test, y_pred_tree, digits=4))\n",
        "\n",
        "# Imprimir área sob curva\n",
        "print('AUC: {:.4f}\\n'.format(roc_auc_score(y_test, y_pred_tree)))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0     0.9995    0.9999    0.9997     71079\n",
            "           1     0.9368    0.7236    0.8165       123\n",
            "\n",
            "    accuracy                         0.9994     71202\n",
            "   macro avg     0.9682    0.8617    0.9081     71202\n",
            "weighted avg     0.9994    0.9994    0.9994     71202\n",
            "\n",
            "AUC: 0.8617\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pOBjxZEH4ehQ"
      },
      "source": [
        "### Construção do Modelo (RUS)\n",
        "\n",
        "Aqui nosso modelo será treinado com dados balanceados pelo método (RUS) e depois suas previsões serão efetuadascom base nos dados de teste.\n",
        "\n",
        "Vamos ver se esse modelo se sai melhor que o modelo de RL (RUS)."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2J-bbdjr4pH7"
      },
      "source": [
        "# Escolher os hyperparameters \n",
        "model_tree_rus = DecisionTreeClassifier(max_depth=8, criterion='entropy')\n",
        "\n",
        "# Fit do Modelo (Treinar)\n",
        "model_tree_rus.fit(X_rus, y_rus)\n",
        "\n",
        "# Fazer previsões em cima dos dados de teste\n",
        "y_pred_tree_rus = model_tree_rus.predict(X_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "L73qyuBs6PbX"
      },
      "source": [
        "#### Avaliando o desempenho do modelo\n",
        "\n",
        "Vamos verificar como nosso primeiro modelo de árvore de decisão, **com dados balanceados**, se comportou:\n",
        "\n",
        "* A quantidade de falsos negativos diminuiu **consideravelmente** (de 34 para 13), o que é um ponto positivo, mas infelizmente o modelo de RL obteve um desempenho superior ao treinar com esses dados. \n",
        "\n",
        "* Houve uma melhora na previsão de fraudes (**de 89 para 110**) e isso gerou um grande aumento nos falsos positivos (**de 6 para 6863**), isso foi um tanto quanto inesperado, esse aumento se torna desastroso quando comparado ao modelo de RL.\n",
        "\n",
        "* Podemos afirmar que o **AUC** de **0.8989**, foi o menor quando comparado aos modelos treinados com dados balanceados.  \n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 499
        },
        "id": "jvMtwxYX6Xdp",
        "outputId": "7d797018-e7d1-4c6e-f17d-1fef13b78bf2"
      },
      "source": [
        "# Plotar matriz de confusão \n",
        "skplt.metrics.plot_confusion_matrix(y_test, y_pred_tree_rus)\n",
        "\n",
        "# Imprimir relatório de classificação\n",
        "print('Relatório de Classificação: \\n', classification_report(y_test, y_pred_tree_rus, digits=4))\n",
        "\n",
        "# Imprimir a Área sob Curva\n",
        "print('AUC: {:.4f}\\n'.format(roc_auc_score(y_test, y_pred_tree_rus)))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Relatório de Classificação: \n",
            "               precision    recall  f1-score   support\n",
            "\n",
            "           0     0.9998    0.9034    0.9492     71079\n",
            "           1     0.0158    0.8943    0.0310       123\n",
            "\n",
            "    accuracy                         0.9034     71202\n",
            "   macro avg     0.5078    0.8989    0.4901     71202\n",
            "weighted avg     0.9981    0.9034    0.9476     71202\n",
            "\n",
            "AUC: 0.8989\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "W2q0P9tZ70uX"
      },
      "source": [
        "### Construção do Modelo (SMOTE)\n",
        "\n",
        "Neste caso, este modelo será treinado com dados balanceados pelo método (SMOTE) e depois suas previsões serão efetuadas com base nos dados de teste."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Enpw4Y5t76ds"
      },
      "source": [
        "# Escolher os Hyperparameters\n",
        "model_tree_smo = DecisionTreeClassifier(max_depth=6, criterion='entropy')\n",
        "\n",
        "# Fit do Modelo(Treinar)\n",
        "model_tree_smo.fit(X_smo, y_smo)\n",
        "\n",
        "# Fazer previsões em cima dos dados de teste\n",
        "y_pred_tree_smo = model_tree_smo.predict(X_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CZLKR_QGPgA8"
      },
      "source": [
        "#### Avaliando o Desempenho do Modelo\n",
        "\n",
        "Com este modelo, podemos ver que:\n",
        "\n",
        "* A quantidade de falsos negativos teve um aumento, quando comparado ao modelo de árvore (RUS), **foram 13 falsos negativos para 18**.\n",
        "\n",
        "* A quantidade de fraudes previstas caiu , **foram 111 contra 113 do modelo de árvore (RUS)**.\n",
        "\n",
        "* Porém, esse modelo de árvore foi o que menos apresentou quantidades de falsos positivos (**são 3588 contra os 6863 do modelo anterior**), uma diminuição de **mais de 3200 casos**, fazendo com que os clientes do banco que escolhesse utilizar esse modelo, tivessem seus cartões bloqueados, menos vezes.\n",
        "\n",
        "* Esse modelo teve um excelente **AUC** mas não o melhor - **0.9016**\n",
        "\n",
        "Os modelos de RL ainda aparentam ter um desempenho melhor na hora de detectar fraudes."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 499
        },
        "id": "EAhn7XHaPmG4",
        "outputId": "c9ef291d-f119-4018-d68e-bd73ce1d2b53"
      },
      "source": [
        "# Plotar matriz de confusão\n",
        "skplt.metrics.plot_confusion_matrix(y_test, y_pred_tree_smo)\n",
        "\n",
        "# Imprimir classification report\n",
        "print('Relatório de Classificação: \\n', classification_report(y_test, y_pred_tree_smo, digits=4))\n",
        "\n",
        "# Imprimir área sob curva \n",
        "print('AUC: {:.4f}\\n'.format(roc_auc_score(y_test,y_pred_tree_smo)))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Relatório de Classificação: \n",
            "               precision    recall  f1-score   support\n",
            "\n",
            "           0     0.9997    0.9495    0.9740     71079\n",
            "           1     0.0284    0.8537    0.0550       123\n",
            "\n",
            "    accuracy                         0.9494     71202\n",
            "   macro avg     0.5141    0.9016    0.5145     71202\n",
            "weighted avg     0.9981    0.9494    0.9724     71202\n",
            "\n",
            "AUC: 0.9016\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UhKop5HER1WQ"
      },
      "source": [
        "### Construção do Modelo (ADASYN)\n",
        "\n",
        "Neste caso, este modelo será treinado com dados balanceados pelo método (ADASYN) e depois suas previsões serão efetuadas como base nos dados de teste."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bGf7jcL3R7yQ"
      },
      "source": [
        "# Escolher os Hyperparameters\n",
        "model_tree_ada = DecisionTreeClassifier(max_depth=8, criterion='entropy')\n",
        "\n",
        "# Fit do Modelo (Treinar)\n",
        "model_tree_ada.fit(X_ada, y_ada)\n",
        "\n",
        "# Fazer previsões em cima dos dados de teste\n",
        "y_pred_tree_ada = model_tree_ada.predict(X_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vTTwoh4rSdR6"
      },
      "source": [
        "#### Avaliando o Desempenho do Modelo\n",
        "\n",
        "No último modelo de Árvore de Decisão, temos os resultados mais equilibrados, mas não os melhores, vamos dar uma olhada:\n",
        "\n",
        "* A quantidade de falsos negativos não foi a menor e nem a maior entre os modelos de árvore, estando no meio **15 falsos negativos**, contra as **13 do modelo de árvore (RUS)**.\n",
        "\n",
        "* A quantidade de fraudes se manteve equilibrada mais uma vez, foram **108 fraudes previstas**, enquanto outro modelo foi capaz de realizar **110 previsões**.\n",
        "\n",
        "* Assim como o modelo (ADASYN) de RL, o modelo de árvore também tem um ponto negativo quanto a quantidade de falsos positivos (sendo **4032 contra os 3588 do SMOTE**), mas vale lembrar que esse não é o modelo com a maior quantidade de falsos positivos. \n",
        "\n",
        "* Esse modelo teve o maior **AUC** dos modelos de árvore - **0.9107**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 499
        },
        "id": "zjJT8oh4SciZ",
        "outputId": "9dbccf47-9a45-4133-b8bd-bc410f1b61c3"
      },
      "source": [
        "# Plotar Matriz de confusão\n",
        "skplt.metrics.plot_confusion_matrix(y_test, y_pred_tree_ada)\n",
        "\n",
        "# Imprimir classification report\n",
        "print('Relatório de classificação: \\n', classification_report(y_test, y_pred_tree_ada, digits=4))\n",
        "\n",
        "# Imprimir área sob curva\n",
        "print('AUC: {:.4f}\\n'.format(roc_auc_score(y_test, y_pred_tree_ada)))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Relatório de classificação: \n",
            "               precision    recall  f1-score   support\n",
            "\n",
            "           0     0.9998    0.9433    0.9707     71079\n",
            "           1     0.0261    0.8780    0.0507       123\n",
            "\n",
            "    accuracy                         0.9432     71202\n",
            "   macro avg     0.5129    0.9107    0.5107     71202\n",
            "weighted avg     0.9981    0.9432    0.9691     71202\n",
            "\n",
            "AUC: 0.9107\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0Bj7zRZMSfO7"
      },
      "source": [
        "## Conclusão\n",
        "\n",
        "Como podemos ver, esse é um projeto diferente dos outros, uma vez que não possui valores ausentes e não foi nescessário fazer um limpeza dos dados. Apesar de termos trabalhado com dados de qualidade e bem limpos, foi necessário lidar com o desbalanceamento e a Transformação PCA, assim demandando um trabalho considerável.\n",
        "\n",
        "É possível concluir que:\n",
        "\n",
        "* Algoritmos de Regressão Logística possuem um desempenho melhor ao lidar com **dados balanceados**.\n",
        "\n",
        "* Algoritmos de Árvore de Decisão foram superiores ao lidar com **dados desbalanceados**.\n",
        "\n",
        "* A solução ideal é aquela que melhor atende a instituição, podendo ser a com o maior **AUC** ou a com o maior número de detecção de fraudes.\n",
        "\n",
        "* O algoritmo que melhor conseguiu prever fraudes foi o **Modelo de Regressão Logística** que treinou com dados balanceados pelo método **ADASYN**."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7VzbTcrcsN_N"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}