2__Atributos_e_Modelagem.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 5,
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.8.5"
    },
    "colab": {
      "name": "2 - Atributos e Modelagem.ipynb",
      "provenance": [],
      "collapsed_sections": []
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "e9de1cb0"
      },
      "source": [
        "# Introdução\n",
        "\n",
        "\n",
        "Algumas informações dadas pelo TSE:\n",
        "\n",
        "\n",
        "- SITUAÇÃO DO REGISTRO\n",
        "    - Indica a situacao do filiado calculado pela aplicacao:\n",
        "        1. REGULAR;\n",
        "        2. EXCLUIDO;\n",
        "        3. SUB_JUDICE;\n",
        "        4. CANCELADO;\n",
        "        5. DESFILIADO;\n",
        "        6. EM_TRANSFERENCIA;\n",
        "        7. TRANSFERIDO;\n",
        "        8. RECEPCIONADO;\n",
        "        9. ERRO;\n",
        "        10. AGUARDANDO ACEITE DO PARTIDO.\n",
        "\n",
        "\n",
        "- TIPO DO REGISTRO\n",
        "    - Tipo da anotacao.\n",
        "        1. Oficial\n",
        "        2. Interna\n",
        "\n",
        "\n",
        "- MOTIVO DO CANCELAMENTO\n",
        "    1. Judicial;\n",
        "    2. A pedido do eleitor;\n",
        "    3. A pedido do partido;\n",
        "    4. Cancelamento automatico;\n",
        "    5. Cancelamento automatico por inscricao inexistente;\n",
        "    6. Cancelamento automatico a pedido do partido;\n",
        "    7. Cancelamento automatico por falecimento;\n",
        "    8. Cancelamento automatico por suspensao de direitos politicos.\n",
        "\n",
        "\n",
        "- DATA DO PROCESSAMENTO\n",
        "    - Data de processamento diario que o registro participou.\n",
        "\n",
        "\n",
        "- DATA DA DESFILIACAO\n",
        "    - Data da desfiliacao que o partido comunicou. Só vale a partir da proxima relacao oficial.\n",
        "    \n",
        "    \n",
        "- DATA DO CANCELAMENTO\n",
        "    - Data do cancelamento da filiacao. Passa a ser desconsiderado a partir desta data.\n",
        "\n",
        "\n",
        "- DATA DA REGULARIZACAO\n",
        "    - Armazena a data de regulacao do registro de filiacao apos decisao judicial ou da corregedoria-geral eleitoral.\n",
        "\n",
        "---\n"
      ],
      "id": "e9de1cb0"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "072dc5ee"
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "from sklearn.model_selection import train_test_split\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "\n",
        "import unicodedata\n",
        "import re \n",
        "\n",
        "def limpar_texto(texto):\n",
        "    # Se vazio, retorna\n",
        "    if texto is None or len(texto.strip()) == 0:\n",
        "        return texto\n",
        "\n",
        "    # Remover acentos\n",
        "    result = unicodedata.normalize('NFKD', texto).encode('ASCII', 'ignore').decode('ASCII')\n",
        "\n",
        "    # Remover caracteres especiais\n",
        "    lista = '-#?º°ª.:/;~^`[{]}\\\\|!$%\"\\'&*()=+,><\\t\\r\\n…'\n",
        "    for i in range(0, len(lista)):\n",
        "        result = result.replace(lista[i], ' ')\n",
        "\n",
        "    # Transformar multiplos espacos em apenas um\n",
        "    result = re.sub(' +', ' ', result)\n",
        "\n",
        "    # Transformar em LOWER case\n",
        "    result = result.lower()\n",
        "\n",
        "    return result"
      ],
      "id": "072dc5ee",
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "22b60945"
      },
      "source": [
        "# Dados CEAF"
      ],
      "id": "22b60945"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HhA8riaJ1xpX",
        "outputId": "4643ecb6-4add-4208-d457-6df4d87081f1"
      },
      "source": [
        "!pip install pickle5"
      ],
      "id": "HhA8riaJ1xpX",
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting pickle5\n",
            "  Downloading pickle5-0.0.11.tar.gz (132 kB)\n",
            "\u001b[?25l\r\u001b[K     |██▌                             | 10 kB 23.1 MB/s eta 0:00:01\r\u001b[K     |█████                           | 20 kB 25.1 MB/s eta 0:00:01\r\u001b[K     |███████▍                        | 30 kB 29.9 MB/s eta 0:00:01\r\u001b[K     |██████████                      | 40 kB 33.7 MB/s eta 0:00:01\r\u001b[K     |████████████▍                   | 51 kB 35.5 MB/s eta 0:00:01\r\u001b[K     |██████████████▉                 | 61 kB 35.0 MB/s eta 0:00:01\r\u001b[K     |█████████████████▍              | 71 kB 32.9 MB/s eta 0:00:01\r\u001b[K     |███████████████████▉            | 81 kB 34.3 MB/s eta 0:00:01\r\u001b[K     |██████████████████████▎         | 92 kB 31.4 MB/s eta 0:00:01\r\u001b[K     |████████████████████████▉       | 102 kB 31.9 MB/s eta 0:00:01\r\u001b[K     |███████████████████████████▎    | 112 kB 31.9 MB/s eta 0:00:01\r\u001b[K     |█████████████████████████████▊  | 122 kB 31.9 MB/s eta 0:00:01\r\u001b[K     |████████████████████████████████| 132 kB 31.9 MB/s \n",
            "\u001b[?25hBuilding wheels for collected packages: pickle5\n",
            "  Building wheel for pickle5 (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for pickle5: filename=pickle5-0.0.11-cp37-cp37m-linux_x86_64.whl size=219300 sha256=1e5b3765de32db4c4e69cf15f7ff717c6bb771237caac5bf166141bfa03fa34b\n",
            "  Stored in directory: /root/.cache/pip/wheels/7e/6a/00/67136a90d6aca437d806d1d3cedf98106e840c97a3e5188198\n",
            "Successfully built pickle5\n",
            "Installing collected packages: pickle5\n",
            "Successfully installed pickle5-0.0.11\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fbf61f79",
        "outputId": "a00c412f-1a21-4174-b105-6a5c119e954a"
      },
      "source": [
        "import pickle5 as pickle\n",
        "\n",
        "with open('/content/df_ceaf.pickle', \"rb\") as fh:\n",
        "  df_ceaf = pickle.load(fh)\n",
        "# df_ceaf = pd.read_pickle('df_ceaf.pickle')\n",
        "\n",
        "df_ceaf.shape"
      ],
      "id": "fbf61f79",
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(1193, 13)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b2a40e8b"
      },
      "source": [
        "# Remover coluna cpf\n",
        "df_ceaf.drop(columns='cpf', axis=1, inplace=True)"
      ],
      "id": "b2a40e8b",
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b9b1de3b"
      },
      "source": [
        "# Separar em TREINO (80%) e TESTE (20%)\n",
        "df_ceaf_train, df_ceaf_test = train_test_split(df_ceaf, test_size=0.2, random_state=171)"
      ],
      "id": "b9b1de3b",
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5673d52b",
        "outputId": "1b660436-4f21-4403-e51f-2eb8f4a63979"
      },
      "source": [
        "# Criar coluna para rotulo (CEAF = 1)\n",
        "df_ceaf_train['rotulo'] = 1\n",
        "df_ceaf_test['rotulo'] = 1"
      ],
      "id": "5673d52b",
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:1: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  \"\"\"Entry point for launching an IPython kernel.\n",
            "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:2: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  \n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "02205a90",
        "outputId": "517ec2ab-8f01-45ad-cb67-7bee49f9f7f7"
      },
      "source": [
        "df_ceaf_train.shape, df_ceaf_test.shape"
      ],
      "id": "02205a90",
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((954, 13), (239, 13))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cfd1b623"
      },
      "source": [
        ""
      ],
      "id": "cfd1b623",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1e40f28c"
      },
      "source": [
        "# Dados Filiados"
      ],
      "id": "1e40f28c"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1783ee4e",
        "outputId": "9ce9f8da-c2aa-4273-f929-b2ffe4dcb143"
      },
      "source": [
        "with open('/content/df_filiados.pickle', \"rb\") as fh:\n",
        "  df_filiados = pickle.load(fh)\n",
        "# df_filiados = pd.read_pickle('df_filiados.pickle')\n",
        "df_filiados.shape"
      ],
      "id": "1783ee4e",
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(246654, 12)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "c07d8d51"
      },
      "source": [
        "# Separar em TREINO (80%) e TESTE (20%)\n",
        "df_filiados_train, df_filiados_test = train_test_split(df_filiados, test_size=0.2, random_state=171)"
      ],
      "id": "c07d8d51",
      "execution_count": 23,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2193af4b",
        "outputId": "1e36b0dd-890f-49e1-9f75-dcd4cb5c9e0b"
      },
      "source": [
        "# Criar coluna para rotulo (NAO-CEAF = 0)\n",
        "df_filiados_train['rotulo'] = 0\n",
        "df_filiados_test['rotulo'] = 0"
      ],
      "id": "2193af4b",
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:2: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  \n",
            "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:3: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  This is separate from the ipykernel package so we can avoid doing imports until\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "d98e5214",
        "outputId": "0e0909f6-ac1d-4198-d00a-8d7fe59ec8d4"
      },
      "source": [
        "df_filiados_train.shape, df_filiados_test.shape"
      ],
      "id": "d98e5214",
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((197323, 13), (49331, 13))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9624f501"
      },
      "source": [
        ""
      ],
      "id": "9624f501",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "c355fa23"
      },
      "source": [
        "# Juntar dados"
      ],
      "id": "c355fa23"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "da47aafd"
      },
      "source": [
        "# Juntar dados de CEAF e Filiados para TREINO e TESTE\n",
        "df_train = pd.concat([df_ceaf_train, df_filiados_train])\n",
        "df_test = pd.concat([df_ceaf_test, df_filiados_test])"
      ],
      "id": "da47aafd",
      "execution_count": 26,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2664b658"
      },
      "source": [
        ""
      ],
      "id": "2664b658",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "b5abcb66"
      },
      "source": [
        "# Engenharia de atributos"
      ],
      "id": "b5abcb66"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "ade302fb",
        "outputId": "59c8fd42-2fc7-4514-f72c-dec8201d630c"
      },
      "source": [
        "df_train.head(30)"
      ],
      "id": "ade302fb",
      "execution_count": 27,
      "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>numero_da_inscricao</th>\n",
              "      <th>nome_do_filiado</th>\n",
              "      <th>sigla_do_partido</th>\n",
              "      <th>uf</th>\n",
              "      <th>data_da_filiacao</th>\n",
              "      <th>situacao_do_registro</th>\n",
              "      <th>tipo_do_registro</th>\n",
              "      <th>data_do_processamento</th>\n",
              "      <th>data_da_desfiliacao</th>\n",
              "      <th>data_do_cancelamento</th>\n",
              "      <th>data_da_regularizacao</th>\n",
              "      <th>motivo_do_cancelamento</th>\n",
              "      <th>rotulo</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>1138</th>\n",
              "      <td>1.408744e+10</td>\n",
              "      <td>raimundo carlos nascimento</td>\n",
              "      <td>PV</td>\n",
              "      <td>BA</td>\n",
              "      <td>2008-01-23</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>923</th>\n",
              "      <td>2.063773e+10</td>\n",
              "      <td>maria jose de arruda campos</td>\n",
              "      <td>PL</td>\n",
              "      <td>MT</td>\n",
              "      <td>2008-01-08</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>191</th>\n",
              "      <td>1.512443e+10</td>\n",
              "      <td>pedro bezerra filho</td>\n",
              "      <td>MDB</td>\n",
              "      <td>PE</td>\n",
              "      <td>2011-09-09</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2019-10-16</td>\n",
              "      <td>NaT</td>\n",
              "      <td>CANCELAMENTO AUTOMÁTICO</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1005</th>\n",
              "      <td>9.437852e+10</td>\n",
              "      <td>rafael de souza e mello coelho da silva</td>\n",
              "      <td>PSOL</td>\n",
              "      <td>RJ</td>\n",
              "      <td>2015-03-31</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>105</th>\n",
              "      <td>5.811562e+09</td>\n",
              "      <td>elson coelho meireles</td>\n",
              "      <td>PSC</td>\n",
              "      <td>AM</td>\n",
              "      <td>1992-04-10</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2013-04-20</td>\n",
              "      <td>2019-10-15</td>\n",
              "      <td>NaT</td>\n",
              "      <td>CANCELAMENTO AUTOMÁTICO</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>797</th>\n",
              "      <td>8.556594e+10</td>\n",
              "      <td>angelo rodrigo ruth</td>\n",
              "      <td>PSC</td>\n",
              "      <td>PR</td>\n",
              "      <td>2007-09-27</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2011-07-30</td>\n",
              "      <td>2010-05-10</td>\n",
              "      <td>NaT</td>\n",
              "      <td>A PEDIDO DO ELEITOR</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>810</th>\n",
              "      <td>2.515779e+10</td>\n",
              "      <td>carmen lucia sacramento</td>\n",
              "      <td>PT</td>\n",
              "      <td>BA</td>\n",
              "      <td>1983-12-10</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2011-09-30</td>\n",
              "      <td>2011-12-16</td>\n",
              "      <td>NaT</td>\n",
              "      <td>JUDICIAL</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>176</th>\n",
              "      <td>2.314142e+09</td>\n",
              "      <td>ubaldino rezende rodrigues</td>\n",
              "      <td>PDT</td>\n",
              "      <td>MT</td>\n",
              "      <td>2011-09-23</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2015-08-05</td>\n",
              "      <td>2019-10-15</td>\n",
              "      <td>NaT</td>\n",
              "      <td>CANCELAMENTO AUTOMÁTICO</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>735</th>\n",
              "      <td>1.346177e+10</td>\n",
              "      <td>alberto batista da silva</td>\n",
              "      <td>PODE</td>\n",
              "      <td>GO</td>\n",
              "      <td>2007-10-03</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2019-10-16</td>\n",
              "      <td>NaT</td>\n",
              "      <td>CANCELAMENTO AUTOMÁTICO</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>481</th>\n",
              "      <td>3.488875e+10</td>\n",
              "      <td>raimundo dos santos furtado</td>\n",
              "      <td>PV</td>\n",
              "      <td>MA</td>\n",
              "      <td>2011-09-09</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>633</th>\n",
              "      <td>5.328904e+10</td>\n",
              "      <td>mauro cezar dorneles da silva</td>\n",
              "      <td>PSDB</td>\n",
              "      <td>RS</td>\n",
              "      <td>2015-04-01</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1140</th>\n",
              "      <td>1.119117e+11</td>\n",
              "      <td>marcos roberto nalon</td>\n",
              "      <td>DC</td>\n",
              "      <td>MG</td>\n",
              "      <td>1995-12-26</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>793</th>\n",
              "      <td>1.002150e+10</td>\n",
              "      <td>abdon de andrade rocha</td>\n",
              "      <td>PV</td>\n",
              "      <td>MA</td>\n",
              "      <td>2020-03-27</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>2020-04-16</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1163</th>\n",
              "      <td>4.597240e+09</td>\n",
              "      <td>carmem lucia de almeida amaral pimenta</td>\n",
              "      <td>MDB</td>\n",
              "      <td>RJ</td>\n",
              "      <td>1996-03-15</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>564</th>\n",
              "      <td>1.601262e+09</td>\n",
              "      <td>gilberto tristao</td>\n",
              "      <td>CIDADANIA</td>\n",
              "      <td>DF</td>\n",
              "      <td>2003-09-17</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>642</th>\n",
              "      <td>2.354666e+10</td>\n",
              "      <td>adir dos santos soares</td>\n",
              "      <td>MDB</td>\n",
              "      <td>RJ</td>\n",
              "      <td>1981-02-16</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>222</th>\n",
              "      <td>6.770141e+09</td>\n",
              "      <td>marcelo rodrigues vaz da costa</td>\n",
              "      <td>PSDB</td>\n",
              "      <td>DF</td>\n",
              "      <td>1995-09-14</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2009-12-18</td>\n",
              "      <td>NaT</td>\n",
              "      <td>JUDICIAL</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>869</th>\n",
              "      <td>8.077261e+09</td>\n",
              "      <td>ilton sacchetti</td>\n",
              "      <td>MDB</td>\n",
              "      <td>SC</td>\n",
              "      <td>1992-02-25</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2007-08-17</td>\n",
              "      <td>2007-08-17</td>\n",
              "      <td>NaT</td>\n",
              "      <td>A PEDIDO DO ELEITOR</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1017</th>\n",
              "      <td>1.773602e+09</td>\n",
              "      <td>jose luciano rodrigues matias</td>\n",
              "      <td>AVANTE</td>\n",
              "      <td>DF</td>\n",
              "      <td>1993-02-03</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>49</th>\n",
              "      <td>4.931257e+10</td>\n",
              "      <td>luiz carlos bueno de lima</td>\n",
              "      <td>MDB</td>\n",
              "      <td>DF</td>\n",
              "      <td>2005-09-30</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2007-01-08</td>\n",
              "      <td>2019-10-15</td>\n",
              "      <td>NaT</td>\n",
              "      <td>CANCELAMENTO AUTOMÁTICO</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1139</th>\n",
              "      <td>3.712597e+10</td>\n",
              "      <td>cristina sueli medeiros da silveira</td>\n",
              "      <td>CIDADANIA</td>\n",
              "      <td>RJ</td>\n",
              "      <td>2011-10-05</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>895</th>\n",
              "      <td>7.089858e+10</td>\n",
              "      <td>marcia dos santos silva correa</td>\n",
              "      <td>PCDOB</td>\n",
              "      <td>RJ</td>\n",
              "      <td>2011-09-20</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2019-10-15</td>\n",
              "      <td>NaT</td>\n",
              "      <td>CANCELAMENTO AUTOMÁTICO</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>542</th>\n",
              "      <td>4.565423e+08</td>\n",
              "      <td>joao jose pires de carvalho</td>\n",
              "      <td>PTB</td>\n",
              "      <td>RO</td>\n",
              "      <td>1985-06-25</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>2020-04-16</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2010-01-04</td>\n",
              "      <td>NaT</td>\n",
              "      <td>CANCELAMENTO AUTOMÁTICO</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>630</th>\n",
              "      <td>6.161046e+10</td>\n",
              "      <td>regina celia ribeiro</td>\n",
              "      <td>PCDOB</td>\n",
              "      <td>RJ</td>\n",
              "      <td>2007-10-05</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>2020-04-16</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2020-04-16</td>\n",
              "      <td>NaT</td>\n",
              "      <td>CANCELAMENTO AUTOMÁTICO</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>666</th>\n",
              "      <td>3.256413e+10</td>\n",
              "      <td>francisco martins de castro</td>\n",
              "      <td>MDB</td>\n",
              "      <td>RS</td>\n",
              "      <td>2003-02-07</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>242</th>\n",
              "      <td>1.148275e+11</td>\n",
              "      <td>silvio cesar vasconcelos brigido</td>\n",
              "      <td>CIDADANIA</td>\n",
              "      <td>MG</td>\n",
              "      <td>2016-03-20</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>831</th>\n",
              "      <td>1.379841e+09</td>\n",
              "      <td>antonio jose nunes de vasconcelos</td>\n",
              "      <td>PDT</td>\n",
              "      <td>PE</td>\n",
              "      <td>2015-08-13</td>\n",
              "      <td>REGULAR</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>767</th>\n",
              "      <td>3.628600e+09</td>\n",
              "      <td>itamar jose dias junqueira</td>\n",
              "      <td>PSOL</td>\n",
              "      <td>RJ</td>\n",
              "      <td>2005-11-30</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2007-09-28</td>\n",
              "      <td>2007-09-28</td>\n",
              "      <td>NaT</td>\n",
              "      <td>A PEDIDO DO ELEITOR</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1087</th>\n",
              "      <td>2.357644e+10</td>\n",
              "      <td>antonio lima campos junior</td>\n",
              "      <td>PSDB</td>\n",
              "      <td>MA</td>\n",
              "      <td>1995-09-30</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2019-10-15</td>\n",
              "      <td>NaT</td>\n",
              "      <td>CANCELAMENTO AUTOMÁTICO</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>948</th>\n",
              "      <td>3.594639e+10</td>\n",
              "      <td>moacir marcos schuh</td>\n",
              "      <td>PP</td>\n",
              "      <td>RS</td>\n",
              "      <td>2000-09-23</td>\n",
              "      <td>CANCELADO</td>\n",
              "      <td>OFICIAL</td>\n",
              "      <td>NaT</td>\n",
              "      <td>2005-07-21</td>\n",
              "      <td>2019-10-15</td>\n",
              "      <td>NaT</td>\n",
              "      <td>CANCELAMENTO AUTOMÁTICO</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      numero_da_inscricao  ... rotulo\n",
              "1138         1.408744e+10  ...      1\n",
              "923          2.063773e+10  ...      1\n",
              "191          1.512443e+10  ...      1\n",
              "1005         9.437852e+10  ...      1\n",
              "105          5.811562e+09  ...      1\n",
              "797          8.556594e+10  ...      1\n",
              "810          2.515779e+10  ...      1\n",
              "176          2.314142e+09  ...      1\n",
              "735          1.346177e+10  ...      1\n",
              "481          3.488875e+10  ...      1\n",
              "633          5.328904e+10  ...      1\n",
              "1140         1.119117e+11  ...      1\n",
              "793          1.002150e+10  ...      1\n",
              "1163         4.597240e+09  ...      1\n",
              "564          1.601262e+09  ...      1\n",
              "642          2.354666e+10  ...      1\n",
              "222          6.770141e+09  ...      1\n",
              "869          8.077261e+09  ...      1\n",
              "1017         1.773602e+09  ...      1\n",
              "49           4.931257e+10  ...      1\n",
              "1139         3.712597e+10  ...      1\n",
              "895          7.089858e+10  ...      1\n",
              "542          4.565423e+08  ...      1\n",
              "630          6.161046e+10  ...      1\n",
              "666          3.256413e+10  ...      1\n",
              "242          1.148275e+11  ...      1\n",
              "831          1.379841e+09  ...      1\n",
              "767          3.628600e+09  ...      1\n",
              "1087         2.357644e+10  ...      1\n",
              "948          3.594639e+10  ...      1\n",
              "\n",
              "[30 rows x 13 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 27
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "6b56103c",
        "outputId": "a8b803dc-836f-48ca-f6e0-062b6fd6cd14"
      },
      "source": [
        "df_train.columns"
      ],
      "id": "6b56103c",
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Index(['numero_da_inscricao', 'nome_do_filiado', 'sigla_do_partido', 'uf',\n",
              "       'data_da_filiacao', 'situacao_do_registro', 'tipo_do_registro',\n",
              "       'data_do_processamento', 'data_da_desfiliacao', 'data_do_cancelamento',\n",
              "       'data_da_regularizacao', 'motivo_do_cancelamento', 'rotulo'],\n",
              "      dtype='object')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9524674a"
      },
      "source": [
        "colunas_selecionadas = [\n",
        "    # Categoricos\n",
        "    'sigla_do_partido', \n",
        "    'uf',\n",
        "    'situacao_do_registro', 'tipo_do_registro','motivo_do_cancelamento',\n",
        "    \n",
        "    # Datas\n",
        "    'data_da_filiacao', \n",
        "    'data_da_desfiliacao', 'data_do_cancelamento',\n",
        "    \n",
        "    # Rotulo\n",
        "    'rotulo']"
      ],
      "id": "9524674a",
      "execution_count": 29,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8b9718ce"
      },
      "source": [
        "# Manter somente colunas selecionadas\n",
        "df_train = df_train[colunas_selecionadas]"
      ],
      "id": "8b9718ce",
      "execution_count": 30,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "df1f0fda"
      },
      "source": [
        ""
      ],
      "id": "df1f0fda",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cfc99622"
      },
      "source": [
        "## Datas\n",
        "\n",
        "- data_da_filiacao\n",
        "- data_da_desfiliacao\n",
        "- data_do_cancelamento"
      ],
      "id": "cfc99622"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "68bb78cd",
        "outputId": "af65eea1-1d8d-4086-b156-7b93f8e8efa2"
      },
      "source": [
        "df_train.columns"
      ],
      "id": "68bb78cd",
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Index(['sigla_do_partido', 'uf', 'situacao_do_registro', 'tipo_do_registro',\n",
              "       'motivo_do_cancelamento', 'data_da_filiacao', 'data_da_desfiliacao',\n",
              "       'data_do_cancelamento', 'rotulo'],\n",
              "      dtype='object')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 31
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4ce4aa96"
      },
      "source": [
        "def merge_datas_finais(x):\n",
        "    '''\n",
        "    Analisa 'data_do_cancelamento' e 'data_da_desfiliacao'\n",
        "    - Se ambas existirem, pega mais recente\n",
        "    - Se apenas uma existir, retorna ela\n",
        "    - Se nenhuma existir, retorna None\n",
        "    '''\n",
        "    if not isinstance(x['data_do_cancelamento'], pd._libs.tslibs.nattype.NaTType) \\\n",
        "       and not isinstance(x['data_da_desfiliacao'], pd._libs.tslibs.nattype.NaTType): \n",
        "            if x['data_do_cancelamento'] >= x['data_da_desfiliacao']:\n",
        "                return x['data_do_cancelamento']\n",
        "            else:\n",
        "                return x.data_da_desfiliacao\n",
        "    elif isinstance(x['data_do_cancelamento'], pd._libs.tslibs.nattype.NaTType) \\\n",
        "       and not isinstance(x['data_da_desfiliacao'], pd._libs.tslibs.nattype.NaTType): \n",
        "        return x['data_da_desfiliacao']\n",
        "        \n",
        "    elif not isinstance(x['data_do_cancelamento'], pd._libs.tslibs.nattype.NaTType) \\\n",
        "       and isinstance(x['data_da_desfiliacao'], pd._libs.tslibs.nattype.NaTType): \n",
        "        return x['data_do_cancelamento']\n",
        "    \n",
        "    else:\n",
        "        return None\n",
        "    \n",
        "df_train['data_final'] = df_train.apply(lambda x: merge_datas_finais(x), axis=1)"
      ],
      "id": "4ce4aa96",
      "execution_count": 32,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "c0c30ee5"
      },
      "source": [
        "def contar_dias(x):\n",
        "    '''\n",
        "    Conta dias entre 'data_da_filiacao' e a 'data_final'\n",
        "    - Caso data_final seja None, retorna NaN\n",
        "    - Caso nao exista data_da_filiacao, retorna NaN\n",
        "    - Ambos existindo, retorna diferenca em dias\n",
        "    '''\n",
        "    if x['data_final'] is None:\n",
        "        return np.NaN\n",
        "    else:\n",
        "        if isinstance(x['data_da_filiacao'], pd._libs.tslibs.nattype.NaTType):\n",
        "            return np.NaN\n",
        "        else:\n",
        "            delta = x['data_final'] - x['data_da_filiacao']\n",
        "            return delta.days\n",
        "\n",
        "df_train['nr_dias_filiado'] = df_train.apply(lambda x: contar_dias(x), axis=1)"
      ],
      "id": "c0c30ee5",
      "execution_count": 33,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4544dda3"
      },
      "source": [
        "def coluna_nao_nat(data):\n",
        "    '''\n",
        "    Se valor de data for invalido, retorna 0.\n",
        "    Caso seja valido, retorna 1.\n",
        "    '''\n",
        "    if isinstance(data, pd._libs.tslibs.nattype.NaTType):\n",
        "        return 0\n",
        "    else:\n",
        "        return 1\n",
        "\n",
        "df_train['bl_desfiliacao'] = df_train['data_da_desfiliacao'].apply(coluna_nao_nat)\n",
        "df_train['bl_cancelamento'] = df_train['data_do_cancelamento'].apply(coluna_nao_nat)"
      ],
      "id": "4544dda3",
      "execution_count": 35,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1f6c25cb"
      },
      "source": [
        "# Criar mes\n",
        "df_train['mes_filiacao'] = df_train.data_da_filiacao.dt.month\n",
        "df_train['mes_filiacao'].fillna(0, inplace=True)\n",
        "df_train['mes_filiacao'] = df_train['mes_filiacao'].astype('int')\n",
        "\n",
        "# Criar ano\n",
        "df_train['ano_filiacao'] = df_train.data_da_filiacao.dt.year % 2\n",
        "df_train['ano_filiacao'].fillna(0, inplace=True)\n",
        "df_train['ano_filiacao'] = df_train['ano_filiacao'].astype('int')"
      ],
      "id": "1f6c25cb",
      "execution_count": 36,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9c79fe97"
      },
      "source": [
        "# Remover colunas originais\n",
        "df_train.drop(columns=['data_da_filiacao', 'data_da_desfiliacao', 'data_do_cancelamento', 'data_final'], axis=1, inplace=True)"
      ],
      "id": "9c79fe97",
      "execution_count": 37,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "c75390b6",
        "outputId": "c2bd3f8a-b545-4ecf-cb40-d730302d5ba3"
      },
      "source": [
        "df_train.columns"
      ],
      "id": "c75390b6",
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Index(['sigla_do_partido', 'uf', 'situacao_do_registro', 'tipo_do_registro',\n",
              "       'motivo_do_cancelamento', 'rotulo', 'nr_dias_filiado', 'bl_desfiliacao',\n",
              "       'bl_cancelamento', 'mes_filiacao', 'ano_filiacao'],\n",
              "      dtype='object')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 38
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "387c58b0"
      },
      "source": [
        ""
      ],
      "id": "387c58b0",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "de8b7322"
      },
      "source": [
        "## Partido e UF\n",
        "\n",
        "- sigla_do_partido\n",
        "- uf"
      ],
      "id": "de8b7322"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "243b9f80"
      },
      "source": [
        "def ver_diff_rotulo(coluna):\n",
        "    '''\n",
        "    Calcula proporcao de cada valor de coluna em relacao ao total\n",
        "    '''\n",
        "    # Quantidade de registros para cada coluna separado por rotulo\n",
        "    c = df_train.groupby([coluna, 'rotulo']).agg(prop=('rotulo', 'count'))\n",
        "    # Divide pelo total de cada rotulo (TOTAL GERAL)\n",
        "    df_c = pd.DataFrame(c/c.groupby(level=1).sum()).reset_index()\n",
        "    # Fazer merge para subtrair (~= pivotear)\n",
        "    df_d = pd.merge(df_c[df_c.rotulo==1], df_c[df_c.rotulo==0], on=coluna, how='right')\n",
        "    # Obtem diferenca das proporcoes entre rotulo=1 e rotulo=0\n",
        "    df_d['diff'] = df_d.prop_x - df_d.prop_y\n",
        "    # Normalizar 0-1\n",
        "    min_diff = min(df_d['diff'])\n",
        "    maxmin_diff = max(df_d['diff']) - min_diff\n",
        "    df_d['diff'] = df_d['diff'].apply(lambda x: (x - min_diff)/maxmin_diff)\n",
        "    # Ordena\n",
        "    df_d = df_d[[coluna,'diff']].sort_values('diff')\n",
        "    # Plota\n",
        "    sns.set(rc={'figure.figsize':(14.7,8.27)})\n",
        "    sns.barplot(x=coluna, y=\"diff\", data=df_d)\n",
        "    _ = plt.xticks(rotation=90)\n",
        "    \n",
        "    # Retorna DataFrame\n",
        "    return df_d"
      ],
      "id": "243b9f80",
      "execution_count": 46,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 597
        },
        "id": "9a516bde",
        "outputId": "296cdecc-827d-47c8-dfa7-3f8ec86487f8"
      },
      "source": [
        "diff_sigla_do_partido = ver_diff_rotulo('sigla_do_partido')"
      ],
      "id": "9a516bde",
      "execution_count": 47,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1058.4x595.44 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "b2e06d51",
        "outputId": "6eacd09b-b811-41d7-aa62-9665a1aa4ac3"
      },
      "source": [
        "# Gerar dicionario para substituicao de PARTIDO por VALOR\n",
        "diff_sigla_do_partido.index = diff_sigla_do_partido.sigla_do_partido\n",
        "diff_sigla_do_partido = diff_sigla_do_partido['diff']\n",
        "diff_sigla_do_partido.fillna(0, inplace=True)\n",
        "diff_sigla_do_partido = diff_sigla_do_partido.to_dict()\n",
        "diff_sigla_do_partido"
      ],
      "id": "b2e06d51",
      "execution_count": 48,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'AVANTE': 0.4444744174254416,\n",
              " 'CIDADANIA': 0.5883795030985551,\n",
              " 'DC': 0.4919375706246683,\n",
              " 'DEM': 0.7845347549601712,\n",
              " 'MDB': 0.0,\n",
              " 'NOVO': 0.56381344458452,\n",
              " 'PATRIOTA': 0.5090322253228586,\n",
              " 'PCB': 0.0,\n",
              " 'PCDOB': 0.6541957042776877,\n",
              " 'PCO': 0.0,\n",
              " 'PDT': 0.54091515194813,\n",
              " 'PHS': 0.0,\n",
              " 'PL': 0.7048199239758387,\n",
              " 'PMB': 0.48460904971541024,\n",
              " 'PMN': 0.5332202757814344,\n",
              " 'PODE': 0.38853085878725513,\n",
              " 'PP': 0.31504733431899284,\n",
              " 'PROS': 0.5894950134736838,\n",
              " 'PRP': 0.0,\n",
              " 'PRTB': 0.5161619952378171,\n",
              " 'PSB': 0.546998130324893,\n",
              " 'PSC': 0.524619095080733,\n",
              " 'PSD': 0.3913697853184328,\n",
              " 'PSDB': 0.17245008464069722,\n",
              " 'PSL': 1.0,\n",
              " 'PSOL': 0.6358348151626643,\n",
              " 'PSTU': 0.534565683882024,\n",
              " 'PT': 0.9541214095224488,\n",
              " 'PTB': 0.11668436730316341,\n",
              " 'PTC': 0.5066737861269409,\n",
              " 'PV': 0.5703497930375947,\n",
              " 'REDE': 0.5369383895876695,\n",
              " 'REPUBLICANOS': 0.3384534493589,\n",
              " 'SOLIDARIEDADE': 0.4300498870440924,\n",
              " 'UP': 0.0}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 48
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "59702d66"
      },
      "source": [
        "# Substituir sigla_do_partido por VALOR consultando dicionario\n",
        "df_train.sigla_do_partido = df_train.sigla_do_partido.apply(lambda x: diff_sigla_do_partido.get(x,0))"
      ],
      "id": "59702d66",
      "execution_count": 49,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3deabf3d"
      },
      "source": [
        ""
      ],
      "id": "3deabf3d",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 525
        },
        "id": "3c71c216",
        "outputId": "1c2859c2-03d6-4c0b-d155-36bc61d01718"
      },
      "source": [
        "diff_uf = ver_diff_rotulo('uf')"
      ],
      "id": "3c71c216",
      "execution_count": 50,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1058.4x595.44 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "13d7a5e1",
        "outputId": "19ffcb59-2d77-4c24-9852-ae4b7d58ce32"
      },
      "source": [
        "# Gerar dicionario para substituicao de UF por VALOR\n",
        "diff_uf.index = diff_uf.uf\n",
        "diff_uf = diff_uf['diff']\n",
        "diff_uf.fillna(0, inplace=True)\n",
        "diff_uf = diff_uf.to_dict()\n",
        "diff_uf"
      ],
      "id": "13d7a5e1",
      "execution_count": 51,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'AC': 0.5060137676898384,\n",
              " 'AL': 0.5339949490570423,\n",
              " 'AM': 0.610175634731232,\n",
              " 'AP': 0.7260694598448074,\n",
              " 'BA': 0.38642886400979665,\n",
              " 'CE': 0.49454926532558374,\n",
              " 'DF': 0.9080857380437889,\n",
              " 'ES': 0.48124426911693846,\n",
              " 'GO': 0.3758391047196174,\n",
              " 'MA': 0.5582461750907729,\n",
              " 'MG': 0.10666089346149399,\n",
              " 'MS': 0.605181599262398,\n",
              " 'MT': 0.6632149011199957,\n",
              " 'PA': 0.6155699205930172,\n",
              " 'PB': 0.5003524373471165,\n",
              " 'PE': 0.5167306593857136,\n",
              " 'PI': 0.4815509495531719,\n",
              " 'PR': 0.43916069077403713,\n",
              " 'RJ': 1.0,\n",
              " 'RN': 0.502036728200881,\n",
              " 'RO': 0.6481946771347766,\n",
              " 'RR': 0.6552807824675334,\n",
              " 'RS': 0.20550692409771504,\n",
              " 'SC': 0.37956229062417335,\n",
              " 'SE': 0.49549230371289577,\n",
              " 'SP': 0.0,\n",
              " 'TO': 0.5056980402598183,\n",
              " 'ZZ': 0.0}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 51
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "27cb81ec"
      },
      "source": [
        "# Substituir uf por VALOR consultando dicionario\n",
        "df_train.uf = df_train.uf.apply(lambda x: diff_uf.get(x,0))"
      ],
      "id": "27cb81ec",
      "execution_count": 52,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "23e2f2d2"
      },
      "source": [
        ""
      ],
      "id": "23e2f2d2",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "c6ef2f74"
      },
      "source": [
        "## Restantes\n",
        "\n",
        "- situacao_do_registro\n",
        "- tipo_do_registro\n",
        "- motivo_do_cancelamento"
      ],
      "id": "c6ef2f74"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "41d4efa0"
      },
      "source": [
        "colunas_dummy = ['situacao_do_registro', 'tipo_do_registro', 'motivo_do_cancelamento']"
      ],
      "id": "41d4efa0",
      "execution_count": 53,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "04bf648a"
      },
      "source": [
        "# Limpar valores das colunas dummy: Virarao nomes das colunas\n",
        "for col in colunas_dummy:\n",
        "    df_train[col] = df_train[col].apply(lambda x: limpar_texto(str(x)).replace(\" \", \"_\"))"
      ],
      "id": "04bf648a",
      "execution_count": 54,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b7039bcd"
      },
      "source": [
        "# Obter one-hot encoding SOMENTE para colunas dummy\n",
        "df_dummy = pd.get_dummies(df_train[colunas_dummy])"
      ],
      "id": "b7039bcd",
      "execution_count": 55,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0bb318f2",
        "outputId": "fd0cef81-9162-4976-a39a-71f2272ea635"
      },
      "source": [
        "df_dummy.shape"
      ],
      "id": "0bb318f2",
      "execution_count": 56,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(198277, 10)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 56
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 224
        },
        "id": "23ca5dc7",
        "outputId": "b5908c59-2cc8-49ff-f70e-fd7e80f91288"
      },
      "source": [
        "df_dummy.head()"
      ],
      "id": "23ca5dc7",
      "execution_count": 57,
      "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>situacao_do_registro_cancelado</th>\n",
              "      <th>situacao_do_registro_desfiliado</th>\n",
              "      <th>situacao_do_registro_regular</th>\n",
              "      <th>situacao_do_registro_sub_judice</th>\n",
              "      <th>tipo_do_registro_oficial</th>\n",
              "      <th>motivo_do_cancelamento_a_pedido_do_eleitor</th>\n",
              "      <th>motivo_do_cancelamento_a_pedido_do_partido</th>\n",
              "      <th>motivo_do_cancelamento_cancelamento_automatico</th>\n",
              "      <th>motivo_do_cancelamento_judicial</th>\n",
              "      <th>motivo_do_cancelamento_nan</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>1138</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>923</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>191</th>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1005</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>105</th>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      situacao_do_registro_cancelado  ...  motivo_do_cancelamento_nan\n",
              "1138                               0  ...                           1\n",
              "923                                0  ...                           1\n",
              "191                                1  ...                           0\n",
              "1005                               0  ...                           1\n",
              "105                                1  ...                           0\n",
              "\n",
              "[5 rows x 10 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 57
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "25b383fa"
      },
      "source": [
        "# Juntar colunas_dummy com colunas restantes\n",
        "colunas_nao_dummy = list(set(df_train.columns) - set(colunas_dummy))\n",
        "df_train = pd.concat([df_dummy, df_train[colunas_nao_dummy]], axis=1)"
      ],
      "id": "25b383fa",
      "execution_count": 58,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8b89a894",
        "outputId": "7f90a6e9-2f20-4891-fb80-fcddd121fd87"
      },
      "source": [
        "df_train.shape"
      ],
      "id": "8b89a894",
      "execution_count": 59,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(198277, 18)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 59
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 224
        },
        "id": "b4a527fe",
        "outputId": "79a05822-a788-4d56-c053-f4e55ae553a3"
      },
      "source": [
        "df_train.head()"
      ],
      "id": "b4a527fe",
      "execution_count": 60,
      "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>situacao_do_registro_cancelado</th>\n",
              "      <th>situacao_do_registro_desfiliado</th>\n",
              "      <th>situacao_do_registro_regular</th>\n",
              "      <th>situacao_do_registro_sub_judice</th>\n",
              "      <th>tipo_do_registro_oficial</th>\n",
              "      <th>motivo_do_cancelamento_a_pedido_do_eleitor</th>\n",
              "      <th>motivo_do_cancelamento_a_pedido_do_partido</th>\n",
              "      <th>motivo_do_cancelamento_cancelamento_automatico</th>\n",
              "      <th>motivo_do_cancelamento_judicial</th>\n",
              "      <th>motivo_do_cancelamento_nan</th>\n",
              "      <th>mes_filiacao</th>\n",
              "      <th>nr_dias_filiado</th>\n",
              "      <th>ano_filiacao</th>\n",
              "      <th>sigla_do_partido</th>\n",
              "      <th>bl_desfiliacao</th>\n",
              "      <th>bl_cancelamento</th>\n",
              "      <th>rotulo</th>\n",
              "      <th>uf</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>1138</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0</td>\n",
              "      <td>0.570350</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0.386429</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>923</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0</td>\n",
              "      <td>0.704820</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0.663215</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>191</th>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>9</td>\n",
              "      <td>2959.0</td>\n",
              "      <td>1</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0.516731</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1005</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>3</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1</td>\n",
              "      <td>0.635835</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>1.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>105</th>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>4</td>\n",
              "      <td>10049.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.524619</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0.610176</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      situacao_do_registro_cancelado  ...        uf\n",
              "1138                               0  ...  0.386429\n",
              "923                                0  ...  0.663215\n",
              "191                                1  ...  0.516731\n",
              "1005                               0  ...  1.000000\n",
              "105                                1  ...  0.610176\n",
              "\n",
              "[5 rows x 18 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 60
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "c14440f7",
        "outputId": "d7031c08-ca38-4a44-fae8-63efbc185fe0"
      },
      "source": [
        "df_train.info()"
      ],
      "id": "c14440f7",
      "execution_count": 61,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "Int64Index: 198277 entries, 1138 to 23327028\n",
            "Data columns (total 18 columns):\n",
            " #   Column                                          Non-Null Count   Dtype  \n",
            "---  ------                                          --------------   -----  \n",
            " 0   situacao_do_registro_cancelado                  198277 non-null  uint8  \n",
            " 1   situacao_do_registro_desfiliado                 198277 non-null  uint8  \n",
            " 2   situacao_do_registro_regular                    198277 non-null  uint8  \n",
            " 3   situacao_do_registro_sub_judice                 198277 non-null  uint8  \n",
            " 4   tipo_do_registro_oficial                        198277 non-null  uint8  \n",
            " 5   motivo_do_cancelamento_a_pedido_do_eleitor      198277 non-null  uint8  \n",
            " 6   motivo_do_cancelamento_a_pedido_do_partido      198277 non-null  uint8  \n",
            " 7   motivo_do_cancelamento_cancelamento_automatico  198277 non-null  uint8  \n",
            " 8   motivo_do_cancelamento_judicial                 198277 non-null  uint8  \n",
            " 9   motivo_do_cancelamento_nan                      198277 non-null  uint8  \n",
            " 10  mes_filiacao                                    198277 non-null  int64  \n",
            " 11  nr_dias_filiado                                 59301 non-null   float64\n",
            " 12  ano_filiacao                                    198277 non-null  int64  \n",
            " 13  sigla_do_partido                                198277 non-null  float64\n",
            " 14  bl_desfiliacao                                  198277 non-null  int64  \n",
            " 15  bl_cancelamento                                 198277 non-null  int64  \n",
            " 16  rotulo                                          198277 non-null  int64  \n",
            " 17  uf                                              198277 non-null  float64\n",
            "dtypes: float64(3), int64(5), uint8(10)\n",
            "memory usage: 20.5 MB\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "813b6217"
      },
      "source": [
        "## Imputar valores em nr_dias_filiado"
      ],
      "id": "813b6217"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "484245ee"
      },
      "source": [
        "from sklearn.experimental import enable_iterative_imputer\n",
        "from sklearn.impute import IterativeImputer\n",
        "\n",
        "# Aprender imputacao a partir dos dados de TREINO\n",
        "impute_dias = IterativeImputer(max_iter=10, random_state=0)\n",
        "impute_dias.fit(df_train)\n",
        "\n",
        "# Imputar dados de TREINO\n",
        "transformed = impute_dias.transform(df_train)"
      ],
      "id": "484245ee",
      "execution_count": 63,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "b9c43e2c",
        "outputId": "9b7140e4-0d49-4d03-9f7f-58d07f1e4722"
      },
      "source": [
        "# Gerar DataFrame com dados imputados\n",
        "df_train = pd.DataFrame(data=transformed, columns=df_train.columns)\n",
        "df_train.info()"
      ],
      "id": "b9c43e2c",
      "execution_count": 64,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 198277 entries, 0 to 198276\n",
            "Data columns (total 18 columns):\n",
            " #   Column                                          Non-Null Count   Dtype  \n",
            "---  ------                                          --------------   -----  \n",
            " 0   situacao_do_registro_cancelado                  198277 non-null  float64\n",
            " 1   situacao_do_registro_desfiliado                 198277 non-null  float64\n",
            " 2   situacao_do_registro_regular                    198277 non-null  float64\n",
            " 3   situacao_do_registro_sub_judice                 198277 non-null  float64\n",
            " 4   tipo_do_registro_oficial                        198277 non-null  float64\n",
            " 5   motivo_do_cancelamento_a_pedido_do_eleitor      198277 non-null  float64\n",
            " 6   motivo_do_cancelamento_a_pedido_do_partido      198277 non-null  float64\n",
            " 7   motivo_do_cancelamento_cancelamento_automatico  198277 non-null  float64\n",
            " 8   motivo_do_cancelamento_judicial                 198277 non-null  float64\n",
            " 9   motivo_do_cancelamento_nan                      198277 non-null  float64\n",
            " 10  mes_filiacao                                    198277 non-null  float64\n",
            " 11  nr_dias_filiado                                 198277 non-null  float64\n",
            " 12  ano_filiacao                                    198277 non-null  float64\n",
            " 13  sigla_do_partido                                198277 non-null  float64\n",
            " 14  bl_desfiliacao                                  198277 non-null  float64\n",
            " 15  bl_cancelamento                                 198277 non-null  float64\n",
            " 16  rotulo                                          198277 non-null  float64\n",
            " 17  uf                                              198277 non-null  float64\n",
            "dtypes: float64(18)\n",
            "memory usage: 27.2 MB\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "80f54b93",
        "outputId": "79b7dc95-e932-4120-ca6c-876ea25b62c5"
      },
      "source": [
        "df_train.nr_dias_filiado.describe()"
      ],
      "id": "80f54b93",
      "execution_count": 65,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "count    198277.000000\n",
              "mean       4219.338732\n",
              "std        2799.292838\n",
              "min       -5562.000000\n",
              "25%        3257.925942\n",
              "50%        3951.681343\n",
              "75%        4598.976176\n",
              "max       75617.000000\n",
              "Name: nr_dias_filiado, dtype: float64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 65
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "68a32e50"
      },
      "source": [
        "# NORMALIZAR\n",
        "min_norm_nr_dias = df_train.nr_dias_filiado.min()\n",
        "max_norm_nr_dias = df_train.nr_dias_filiado.max()\n",
        "\n",
        "def normalizar_nr_dias(coluna):\n",
        "    return (coluna - min_norm_nr_dias) / (max_norm_nr_dias-min_norm_nr_dias)\n",
        "\n",
        "df_train.nr_dias_filiado = df_train.nr_dias_filiado.apply(normalizar_nr_dias)"
      ],
      "id": "68a32e50",
      "execution_count": 66,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0a964766",
        "outputId": "c2cc6bc0-bef0-4d8e-abd4-108da9122278"
      },
      "source": [
        "df_train.nr_dias_filiado.describe()"
      ],
      "id": "0a964766",
      "execution_count": 67,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "count    198277.000000\n",
              "mean          0.120491\n",
              "std           0.034483\n",
              "min           0.000000\n",
              "25%           0.108648\n",
              "50%           0.117194\n",
              "75%           0.125168\n",
              "max           1.000000\n",
              "Name: nr_dias_filiado, dtype: float64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 67
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "338a84f9"
      },
      "source": [
        "# Selecionar atributos"
      ],
      "id": "338a84f9"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 224
        },
        "id": "eccedc7d",
        "outputId": "88ece5bf-0401-457a-f687-df471eeff499"
      },
      "source": [
        "df_train.head()"
      ],
      "id": "eccedc7d",
      "execution_count": 68,
      "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>situacao_do_registro_cancelado</th>\n",
              "      <th>situacao_do_registro_desfiliado</th>\n",
              "      <th>situacao_do_registro_regular</th>\n",
              "      <th>situacao_do_registro_sub_judice</th>\n",
              "      <th>tipo_do_registro_oficial</th>\n",
              "      <th>motivo_do_cancelamento_a_pedido_do_eleitor</th>\n",
              "      <th>motivo_do_cancelamento_a_pedido_do_partido</th>\n",
              "      <th>motivo_do_cancelamento_cancelamento_automatico</th>\n",
              "      <th>motivo_do_cancelamento_judicial</th>\n",
              "      <th>motivo_do_cancelamento_nan</th>\n",
              "      <th>mes_filiacao</th>\n",
              "      <th>nr_dias_filiado</th>\n",
              "      <th>ano_filiacao</th>\n",
              "      <th>sigla_do_partido</th>\n",
              "      <th>bl_desfiliacao</th>\n",
              "      <th>bl_cancelamento</th>\n",
              "      <th>rotulo</th>\n",
              "      <th>uf</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.125196</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.570350</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.386429</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.120311</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.704820</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.663215</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>9.0</td>\n",
              "      <td>0.104966</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.516731</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.113845</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.635835</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>0.192303</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.524619</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.610176</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   situacao_do_registro_cancelado  ...        uf\n",
              "0                             0.0  ...  0.386429\n",
              "1                             0.0  ...  0.663215\n",
              "2                             1.0  ...  0.516731\n",
              "3                             0.0  ...  1.000000\n",
              "4                             1.0  ...  0.610176\n",
              "\n",
              "[5 rows x 18 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 68
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "df5b866c"
      },
      "source": [
        "from sklearn.feature_selection import SelectKBest\n",
        "from sklearn.feature_selection import chi2\n",
        "\n",
        "fs_chi2 = SelectKBest(score_func=chi2, k='all')\n",
        "\n",
        "# Analise de Atributos nos dados de Treino\n",
        "fs_chi2.fit(df_train.loc[:,df_train.columns != 'rotulo'], df_train.rotulo)"
      ],
      "id": "df5b866c",
      "execution_count": 69,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "da030c2c",
        "outputId": "0e6b5a4e-2c84-4eb3-e52c-cb8a6c385f1b"
      },
      "source": [
        "from matplotlib import pyplot\n",
        "\n",
        "scores = [fs_chi2.scores_[i] for i in range(len(fs_chi2.scores_))]\n",
        "colunas = [df_train.columns[1:][i] for i in range(len(fs_chi2.scores_))]\n",
        "\n",
        "colunas_finais = []\n",
        "\n",
        "for i, score in enumerate(scores):\n",
        "    if score > 0.05:\n",
        "        print('Atributo: %s: %f -- IN' % (colunas[i], score))\n",
        "        colunas_finais.append(colunas[i])\n",
        "    else:\n",
        "        print('Atributo: %s: %f -- OUT' % (colunas[i], score))\n",
        "\n",
        "# plot the scores\n",
        "pyplot.bar(colunas, scores)\n",
        "pyplot.xticks(rotation=90)\n",
        "pyplot.show()"
      ],
      "id": "da030c2c",
      "execution_count": 84,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Atributo: situacao_do_registro_desfiliado: 20.716242 -- IN\n",
            "Atributo: situacao_do_registro_regular: 5.653615 -- IN\n",
            "Atributo: situacao_do_registro_sub_judice: 13.831126 -- IN\n",
            "Atributo: tipo_do_registro_oficial: 0.048347 -- OUT\n",
            "Atributo: motivo_do_cancelamento_a_pedido_do_eleitor: 0.000000 -- OUT\n",
            "Atributo: motivo_do_cancelamento_a_pedido_do_partido: 15.009217 -- IN\n",
            "Atributo: motivo_do_cancelamento_cancelamento_automatico: 2.427026 -- IN\n",
            "Atributo: motivo_do_cancelamento_judicial: 79.205195 -- IN\n",
            "Atributo: motivo_do_cancelamento_nan: 1.662131 -- IN\n",
            "Atributo: mes_filiacao: 19.000368 -- IN\n",
            "Atributo: nr_dias_filiado: 0.276492 -- IN\n",
            "Atributo: ano_filiacao: 0.003351 -- OUT\n",
            "Atributo: sigla_do_partido: 0.070024 -- IN\n",
            "Atributo: bl_desfiliacao: 7.981234 -- IN\n",
            "Atributo: bl_cancelamento: 33.343051 -- IN\n",
            "Atributo: rotulo: 39.939067 -- IN\n",
            "Atributo: uf: 89.343190 -- IN\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1058.4x595.44 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xu7g-7ZH7wou",
        "outputId": "d10c1fb1-17cd-4429-868f-2443dcd618f2"
      },
      "source": [
        "colunas_finais"
      ],
      "id": "xu7g-7ZH7wou",
      "execution_count": 85,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['situacao_do_registro_desfiliado',\n",
              " 'situacao_do_registro_regular',\n",
              " 'situacao_do_registro_sub_judice',\n",
              " 'motivo_do_cancelamento_a_pedido_do_partido',\n",
              " 'motivo_do_cancelamento_cancelamento_automatico',\n",
              " 'motivo_do_cancelamento_judicial',\n",
              " 'motivo_do_cancelamento_nan',\n",
              " 'mes_filiacao',\n",
              " 'nr_dias_filiado',\n",
              " 'sigla_do_partido',\n",
              " 'bl_desfiliacao',\n",
              " 'bl_cancelamento',\n",
              " 'rotulo',\n",
              " 'uf']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 85
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 790
        },
        "id": "57c711f7",
        "outputId": "9cc02611-fc5b-4105-ca11-807f73c47e79"
      },
      "source": [
        "# Matrix de Correlacao (valores absolutos)\n",
        "corr_matrix = df_train[colunas_finais].corr().abs()\n",
        "\n",
        "# Plotar\n",
        "f, ax = plt.subplots(figsize=(10, 8))\n",
        "sns.heatmap(corr_matrix, mask=np.zeros_like(corr_matrix, dtype=np.bool), cmap=sns.diverging_palette(220, 10, as_cmap=True),\n",
        "            square=True, ax=ax)"
      ],
      "id": "57c711f7",
      "execution_count": 87,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fc64553c410>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 87
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x576 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "153916f4",
        "outputId": "a15688f9-e941-43c6-9631-63308784b7b4"
      },
      "source": [
        "# Limite para correlacao\n",
        "limite = 0.8\n",
        "\n",
        "# Selecionar colunas com correlacao acima do limite\n",
        "upper = corr_matrix.where(np.triu(np.ones(corr_matrix.shape), k=1).astype(np.bool))\n",
        "alta_correlacao = [column for column in upper.columns if any(upper[column] > limite)]\n",
        "alta_correlacao"
      ],
      "id": "153916f4",
      "execution_count": 90,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['bl_cancelamento']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 90
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "90c8749f",
        "outputId": "4077b3c0-ae61-4da8-d76d-211ff5caa102"
      },
      "source": [
        "# Eliminar as de alta_correlacao\n",
        "colunas_finais = list(set(colunas_finais) - set(alta_correlacao))\n",
        "colunas_finais"
      ],
      "id": "90c8749f",
      "execution_count": 92,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['situacao_do_registro_desfiliado',\n",
              " 'mes_filiacao',\n",
              " 'motivo_do_cancelamento_a_pedido_do_partido',\n",
              " 'nr_dias_filiado',\n",
              " 'motivo_do_cancelamento_judicial',\n",
              " 'sigla_do_partido',\n",
              " 'bl_desfiliacao',\n",
              " 'situacao_do_registro_regular',\n",
              " 'situacao_do_registro_sub_judice',\n",
              " 'rotulo',\n",
              " 'motivo_do_cancelamento_cancelamento_automatico',\n",
              " 'uf',\n",
              " 'motivo_do_cancelamento_nan']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 92
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "57a9ce0a"
      },
      "source": [
        "# Filtrar colunas finais\n",
        "df_train = df_train[colunas_finais]"
      ],
      "id": "57a9ce0a",
      "execution_count": 93,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "34cb4728",
        "outputId": "51729f92-7ff6-4065-a70b-a957eb8d4076"
      },
      "source": [
        "df_train.shape"
      ],
      "id": "34cb4728",
      "execution_count": 94,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(198277, 13)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 94
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 224
        },
        "id": "7d1f408f",
        "outputId": "f5605d4b-b018-4448-e203-2f8eca6e652e"
      },
      "source": [
        "df_train.head()"
      ],
      "id": "7d1f408f",
      "execution_count": 95,
      "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>situacao_do_registro_desfiliado</th>\n",
              "      <th>mes_filiacao</th>\n",
              "      <th>motivo_do_cancelamento_a_pedido_do_partido</th>\n",
              "      <th>nr_dias_filiado</th>\n",
              "      <th>motivo_do_cancelamento_judicial</th>\n",
              "      <th>sigla_do_partido</th>\n",
              "      <th>bl_desfiliacao</th>\n",
              "      <th>situacao_do_registro_regular</th>\n",
              "      <th>situacao_do_registro_sub_judice</th>\n",
              "      <th>rotulo</th>\n",
              "      <th>motivo_do_cancelamento_cancelamento_automatico</th>\n",
              "      <th>uf</th>\n",
              "      <th>motivo_do_cancelamento_nan</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.125196</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.570350</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.386429</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.120311</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.704820</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.663215</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.0</td>\n",
              "      <td>9.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.104966</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.516731</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.113845</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.635835</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.192303</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.524619</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.610176</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   situacao_do_registro_desfiliado  ...  motivo_do_cancelamento_nan\n",
              "0                              0.0  ...                         1.0\n",
              "1                              0.0  ...                         1.0\n",
              "2                              0.0  ...                         0.0\n",
              "3                              0.0  ...                         1.0\n",
              "4                              0.0  ...                         0.0\n",
              "\n",
              "[5 rows x 13 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 95
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "34f50d5c"
      },
      "source": [
        ""
      ],
      "id": "34f50d5c",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "63ae957b"
      },
      "source": [
        "# Dados de Teste"
      ],
      "id": "63ae957b"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "a9ac1f5a"
      },
      "source": [
        "df_test = df_test[colunas_selecionadas]\n",
        "df_test['data_final'] = df_test.apply(lambda x: merge_datas_finais(x), axis=1)\n",
        "df_test['nr_dias_filiado'] = df_test.apply(lambda x: contar_dias(x), axis=1)\n",
        "\n",
        "df_test['bl_desfiliacao'] = df_test['data_da_desfiliacao'].apply(coluna_nao_nat)\n",
        "df_test['bl_cancelamento'] = df_test['data_do_cancelamento'].apply(coluna_nao_nat)\n",
        "\n",
        "df_test['mes_filiacao'] = df_test.data_da_filiacao.dt.month\n",
        "df_test['mes_filiacao'].fillna(0, inplace=True)\n",
        "df_test['mes_filiacao'] = df_test['mes_filiacao'].astype('int')\n",
        "\n",
        "df_test['ano_filiacao'] = df_test.data_da_filiacao.dt.year % 2\n",
        "df_test['ano_filiacao'].fillna(0, inplace=True)\n",
        "df_test['ano_filiacao'] = df_test['ano_filiacao'].astype('int')\n",
        "\n",
        "df_test.drop(columns=['data_da_filiacao', 'data_da_desfiliacao', \n",
        "                      'data_do_cancelamento', 'data_final'], axis=1, inplace=True)\n",
        "\n",
        "df_test.sigla_do_partido = df_test.sigla_do_partido.apply(lambda x: diff_sigla_do_partido.get(x,0))\n",
        "df_test.uf = df_test.uf.apply(lambda x: diff_uf.get(x,0))\n",
        "\n",
        "# Limpar valores das colunas dummy: Virarao nomes das colunas\n",
        "for col in colunas_dummy:\n",
        "    df_test[col] = df_test[col].apply(lambda x: limpar_texto(str(x)).replace(\" \", \"_\"))\n",
        "\n",
        "df_dummy_test = pd.get_dummies(df_test[colunas_dummy])\n",
        "\n",
        "df_test = pd.concat([df_dummy_test, df_test[colunas_nao_dummy]], axis=1)\n",
        "\n",
        "transformed_test = impute_dias.transform(df_test)\n",
        "df_test = pd.DataFrame(data=transformed_test, columns=df_test.columns)\n",
        "\n",
        "df_test.nr_dias_filiado = df_test.nr_dias_filiado.apply(normalizar_nr_dias)\n",
        "\n",
        "df_test = df_test[colunas_finais]"
      ],
      "id": "a9ac1f5a",
      "execution_count": 96,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "f863c838",
        "outputId": "915cdcb9-ac11-4daf-dc9e-34a56509c298"
      },
      "source": [
        "df_test.shape"
      ],
      "id": "f863c838",
      "execution_count": 103,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(49570, 13)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 103
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 224
        },
        "id": "ed53916c",
        "outputId": "3befdb20-d8bb-4648-d061-51d27d9c5529"
      },
      "source": [
        "df_test.head()"
      ],
      "id": "ed53916c",
      "execution_count": 147,
      "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>situacao_do_registro_desfiliado</th>\n",
              "      <th>mes_filiacao</th>\n",
              "      <th>motivo_do_cancelamento_a_pedido_do_partido</th>\n",
              "      <th>nr_dias_filiado</th>\n",
              "      <th>motivo_do_cancelamento_judicial</th>\n",
              "      <th>sigla_do_partido</th>\n",
              "      <th>bl_desfiliacao</th>\n",
              "      <th>situacao_do_registro_regular</th>\n",
              "      <th>situacao_do_registro_sub_judice</th>\n",
              "      <th>rotulo</th>\n",
              "      <th>motivo_do_cancelamento_cancelamento_automatico</th>\n",
              "      <th>uf</th>\n",
              "      <th>motivo_do_cancelamento_nan</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.0</td>\n",
              "      <td>9.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.118946</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.338453</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.615570</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.202060</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.172450</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.610176</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.0</td>\n",
              "      <td>5.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.120273</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.546998</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.205507</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.125317</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.388531</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.615570</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.132703</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.116684</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.481551</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   situacao_do_registro_desfiliado  ...  motivo_do_cancelamento_nan\n",
              "0                              0.0  ...                         1.0\n",
              "1                              0.0  ...                         0.0\n",
              "2                              0.0  ...                         1.0\n",
              "3                              0.0  ...                         1.0\n",
              "4                              0.0  ...                         1.0\n",
              "\n",
              "[5 rows x 13 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 147
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "97d67896",
        "outputId": "6dc891b3-2730-446e-f053-693d208fa8d8"
      },
      "source": [
        "df_train.info()"
      ],
      "id": "97d67896",
      "execution_count": 145,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 198277 entries, 0 to 198276\n",
            "Data columns (total 13 columns):\n",
            " #   Column                                          Non-Null Count   Dtype  \n",
            "---  ------                                          --------------   -----  \n",
            " 0   situacao_do_registro_desfiliado                 198277 non-null  float64\n",
            " 1   mes_filiacao                                    198277 non-null  float64\n",
            " 2   motivo_do_cancelamento_a_pedido_do_partido      198277 non-null  float64\n",
            " 3   nr_dias_filiado                                 198277 non-null  float64\n",
            " 4   motivo_do_cancelamento_judicial                 198277 non-null  float64\n",
            " 5   sigla_do_partido                                198277 non-null  float64\n",
            " 6   bl_desfiliacao                                  198277 non-null  float64\n",
            " 7   situacao_do_registro_regular                    198277 non-null  float64\n",
            " 8   situacao_do_registro_sub_judice                 198277 non-null  float64\n",
            " 9   rotulo                                          198277 non-null  float64\n",
            " 10  motivo_do_cancelamento_cancelamento_automatico  198277 non-null  float64\n",
            " 11  uf                                              198277 non-null  float64\n",
            " 12  motivo_do_cancelamento_nan                      198277 non-null  float64\n",
            "dtypes: float64(13)\n",
            "memory usage: 19.7 MB\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "54e97984"
      },
      "source": [
        "# Machine Learning"
      ],
      "id": "54e97984"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "06w-7VxFP1Tu"
      },
      "source": [
        "## Balanceamento dos dados"
      ],
      "id": "06w-7VxFP1Tu"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "388dbdd8",
        "outputId": "e5879a8e-7536-4529-c711-869af4e614c2"
      },
      "source": [
        "features = df_train.columns[df_train.columns != 'rotulo'].values.tolist()\n",
        "features"
      ],
      "id": "388dbdd8",
      "execution_count": 105,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['situacao_do_registro_desfiliado',\n",
              " 'mes_filiacao',\n",
              " 'motivo_do_cancelamento_a_pedido_do_partido',\n",
              " 'nr_dias_filiado',\n",
              " 'motivo_do_cancelamento_judicial',\n",
              " 'sigla_do_partido',\n",
              " 'bl_desfiliacao',\n",
              " 'situacao_do_registro_regular',\n",
              " 'situacao_do_registro_sub_judice',\n",
              " 'motivo_do_cancelamento_cancelamento_automatico',\n",
              " 'uf',\n",
              " 'motivo_do_cancelamento_nan']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 105
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "296f757f"
      },
      "source": [
        "X_train = df_train.loc[:,features].to_numpy()\n",
        "X_test = df_test.loc[:,features].to_numpy()\n",
        "y_train = df_train.rotulo.values.tolist()\n",
        "y_test = df_test.rotulo.values.tolist()"
      ],
      "id": "296f757f",
      "execution_count": 106,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "f4727657",
        "outputId": "77361138-20ee-461f-cafa-a71249d01199"
      },
      "source": [
        "X_train.shape, X_test.shape, len(y_train), len(y_test)"
      ],
      "id": "f4727657",
      "execution_count": 107,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((198277, 12), (49570, 12), 198277, 49570)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 107
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "df2767e0",
        "outputId": "cbcd5cce-abad-4fec-dd0b-84a11fb8a154"
      },
      "source": [
        "from collections import Counter\n",
        "\n",
        "ct_tr = Counter(y_train)\n",
        "ct_te = Counter(y_test)\n",
        "ct_tr, ct_te"
      ],
      "id": "df2767e0",
      "execution_count": 108,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(Counter({0.0: 197323, 1.0: 954}), Counter({0.0: 49331, 1.0: 239}))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 108
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "b372751e",
        "outputId": "6752862d-6294-45ec-e7b9-24029d9642f0"
      },
      "source": [
        "np.random.seed(171)\n",
        "nsmp = Counter(y_train)[0]-Counter(y_train)[1]\n",
        "print(nsmp)\n",
        "indices = np.random.choice(range(Counter(y_train)[1]), size=nsmp, replace=True)\n",
        "X_train = np.concatenate((X_train, X_train[np.array(y_train) == 1][indices]))\n",
        "print(X_train.shape)\n",
        "\n",
        "y_train = y_train + [1]*nsmp\n",
        "Counter(y_train), Counter(y_test)"
      ],
      "id": "b372751e",
      "execution_count": 109,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "196369\n",
            "(394646, 12)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(Counter({0.0: 197323, 1.0: 197323}), Counter({0.0: 49331, 1.0: 239}))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 109
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MclLEATnP4LC"
      },
      "source": [
        "## Treinamento"
      ],
      "id": "MclLEATnP4LC"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "c45078cb"
      },
      "source": [
        "from keras.models import Sequential\n",
        "from keras.layers import LSTM\n",
        "from keras.layers import Dense\n",
        "from keras import metrics\n",
        "from keras.optimizers import Adam\n",
        "from keras.callbacks import EarlyStopping\n",
        "\n",
        "np.random.seed(171)"
      ],
      "id": "c45078cb",
      "execution_count": 110,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9af14244"
      },
      "source": [
        "model = Sequential()\n",
        "model.add(Dense(64, input_dim=len(features), activation='relu'))\n",
        "model.add(Dense(32, activation='relu'))\n",
        "model.add(Dense(2, activation='sigmoid'))\n",
        "\n",
        "opt = Adam(learning_rate=0.001)\n",
        "model.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['accuracy', metrics.Precision(), metrics.Recall(), metrics.AUC()])"
      ],
      "id": "9af14244",
      "execution_count": 171,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "82e00e15",
        "outputId": "0120fd22-72a8-4c80-85d2-f97a22b603a0"
      },
      "source": [
        "print(model.summary())"
      ],
      "id": "82e00e15",
      "execution_count": 172,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_8\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "dense_19 (Dense)             (None, 64)                832       \n",
            "_________________________________________________________________\n",
            "dense_20 (Dense)             (None, 32)                2080      \n",
            "_________________________________________________________________\n",
            "dense_21 (Dense)             (None, 2)                 66        \n",
            "=================================================================\n",
            "Total params: 2,978\n",
            "Trainable params: 2,978\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Xa4I8uQY-ar0",
        "outputId": "82132e61-246f-443a-d180-e48dcb58d731"
      },
      "source": [
        "y_train_k = tf.keras.utils.to_categorical(y_train, num_classes=2)\n",
        "y_test_k = tf.keras.utils.to_categorical(y_test, num_classes=2)\n",
        "\n",
        "X_train_k = X_train.reshape((X_train.shape[0],X_train.shape[1]))\n",
        "X_test_k = X_test.reshape((X_test.shape[0],X_test.shape[1]))\n",
        "\n",
        "y_train_k.shape, y_test_k.shape, X_train_k.shape, X_test_k.shape"
      ],
      "id": "Xa4I8uQY-ar0",
      "execution_count": 173,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((394646, 2), (49570, 2), (394646, 12), (49570, 12))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 173
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "f36893fe",
        "outputId": "d5938f16-02a2-428d-9374-4c670ac12ff5"
      },
      "source": [
        "callb = EarlyStopping(monitor='loss', patience=5, restore_best_weights=True)\n",
        "\n",
        "model.fit(X_train, y_train_k, validation_data=(X_test, y_test_k), epochs=50, initial_epoch=0, batch_size=256, callbacks=[callb])"
      ],
      "id": "f36893fe",
      "execution_count": 174,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "1542/1542 [==============================] - 8s 4ms/step - loss: 0.6441 - accuracy: 0.6278 - precision_5: 0.7439 - recall_5: 0.0712 - auc_5: 0.6184 - val_loss: 0.6479 - val_accuracy: 0.6535 - val_precision_5: 0.5382 - val_recall_5: 0.0946 - val_auc_5: 0.6611\n",
            "Epoch 2/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.6111 - accuracy: 0.6658 - precision_5: 0.7977 - recall_5: 0.1760 - auc_5: 0.7026 - val_loss: 0.5468 - val_accuracy: 0.7393 - val_precision_5: 0.8095 - val_recall_5: 0.2810 - val_auc_5: 0.8001\n",
            "Epoch 3/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.5979 - accuracy: 0.6761 - precision_5: 0.7924 - recall_5: 0.2569 - auc_5: 0.7291 - val_loss: 0.5961 - val_accuracy: 0.6877 - val_precision_5: 0.7737 - val_recall_5: 0.2663 - val_auc_5: 0.7354\n",
            "Epoch 4/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.5842 - accuracy: 0.6861 - precision_5: 0.8033 - recall_5: 0.2914 - auc_5: 0.7435 - val_loss: 0.5077 - val_accuracy: 0.7506 - val_precision_5: 0.8658 - val_recall_5: 0.3581 - val_auc_5: 0.8306\n",
            "Epoch 5/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.5735 - accuracy: 0.6956 - precision_5: 0.8153 - recall_5: 0.2917 - auc_5: 0.7522 - val_loss: 0.6766 - val_accuracy: 0.6001 - val_precision_5: 0.6603 - val_recall_5: 0.2120 - val_auc_5: 0.6360\n",
            "Epoch 6/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.5627 - accuracy: 0.7020 - precision_5: 0.8279 - recall_5: 0.3074 - auc_5: 0.7613 - val_loss: 0.5726 - val_accuracy: 0.7034 - val_precision_5: 0.7939 - val_recall_5: 0.2891 - val_auc_5: 0.7564\n",
            "Epoch 7/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.5405 - accuracy: 0.7205 - precision_5: 0.8570 - recall_5: 0.3008 - auc_5: 0.7813 - val_loss: 0.6088 - val_accuracy: 0.6898 - val_precision_5: 0.7276 - val_recall_5: 0.1915 - val_auc_5: 0.7153\n",
            "Epoch 8/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.5020 - accuracy: 0.7610 - precision_5: 0.8875 - recall_5: 0.2878 - auc_5: 0.8116 - val_loss: 0.4081 - val_accuracy: 0.8368 - val_precision_5: 0.8838 - val_recall_5: 0.4313 - val_auc_5: 0.8891\n",
            "Epoch 9/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.4644 - accuracy: 0.7954 - precision_5: 0.9033 - recall_5: 0.3158 - auc_5: 0.8452 - val_loss: 0.3566 - val_accuracy: 0.8856 - val_precision_5: 0.9350 - val_recall_5: 0.4300 - val_auc_5: 0.9267\n",
            "Epoch 10/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.4227 - accuracy: 0.8243 - precision_5: 0.9179 - recall_5: 0.3747 - auc_5: 0.8773 - val_loss: 0.4460 - val_accuracy: 0.8232 - val_precision_5: 0.8810 - val_recall_5: 0.3542 - val_auc_5: 0.8660\n",
            "Epoch 11/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.3874 - accuracy: 0.8411 - precision_5: 0.9294 - recall_5: 0.4339 - auc_5: 0.8990 - val_loss: 0.4371 - val_accuracy: 0.8086 - val_precision_5: 0.8769 - val_recall_5: 0.4104 - val_auc_5: 0.8713\n",
            "Epoch 12/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.3575 - accuracy: 0.8528 - precision_5: 0.9363 - recall_5: 0.4886 - auc_5: 0.9139 - val_loss: 0.3890 - val_accuracy: 0.8281 - val_precision_5: 0.9251 - val_recall_5: 0.4443 - val_auc_5: 0.9022\n",
            "Epoch 13/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.3352 - accuracy: 0.8594 - precision_5: 0.9426 - recall_5: 0.5240 - auc_5: 0.9230 - val_loss: 0.4124 - val_accuracy: 0.8109 - val_precision_5: 0.9030 - val_recall_5: 0.4924 - val_auc_5: 0.8865\n",
            "Epoch 14/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.3196 - accuracy: 0.8652 - precision_5: 0.9469 - recall_5: 0.5466 - auc_5: 0.9288 - val_loss: 0.3461 - val_accuracy: 0.8399 - val_precision_5: 0.9393 - val_recall_5: 0.5773 - val_auc_5: 0.9229\n",
            "Epoch 15/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.3070 - accuracy: 0.8682 - precision_5: 0.9512 - recall_5: 0.5663 - auc_5: 0.9335 - val_loss: 0.2681 - val_accuracy: 0.8974 - val_precision_5: 0.9678 - val_recall_5: 0.6986 - val_auc_5: 0.9579\n",
            "Epoch 16/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2971 - accuracy: 0.8715 - precision_5: 0.9528 - recall_5: 0.5842 - auc_5: 0.9374 - val_loss: 0.4665 - val_accuracy: 0.7769 - val_precision_5: 0.8964 - val_recall_5: 0.4424 - val_auc_5: 0.8560\n",
            "Epoch 17/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2881 - accuracy: 0.8745 - precision_5: 0.9548 - recall_5: 0.5940 - auc_5: 0.9406 - val_loss: 0.3926 - val_accuracy: 0.8099 - val_precision_5: 0.8985 - val_recall_5: 0.6253 - val_auc_5: 0.9048\n",
            "Epoch 18/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2832 - accuracy: 0.8753 - precision_5: 0.9568 - recall_5: 0.5983 - auc_5: 0.9421 - val_loss: 0.2800 - val_accuracy: 0.8708 - val_precision_5: 0.9590 - val_recall_5: 0.7156 - val_auc_5: 0.9519\n",
            "Epoch 19/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2775 - accuracy: 0.8783 - precision_5: 0.9597 - recall_5: 0.5990 - auc_5: 0.9437 - val_loss: 0.4206 - val_accuracy: 0.7959 - val_precision_5: 0.9129 - val_recall_5: 0.5227 - val_auc_5: 0.8890\n",
            "Epoch 20/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2739 - accuracy: 0.8799 - precision_5: 0.9620 - recall_5: 0.5986 - auc_5: 0.9446 - val_loss: 0.2886 - val_accuracy: 0.8635 - val_precision_5: 0.9489 - val_recall_5: 0.7205 - val_auc_5: 0.9501\n",
            "Epoch 21/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2663 - accuracy: 0.8842 - precision_5: 0.9657 - recall_5: 0.5980 - auc_5: 0.9475 - val_loss: 0.3345 - val_accuracy: 0.8325 - val_precision_5: 0.9475 - val_recall_5: 0.6587 - val_auc_5: 0.9310\n",
            "Epoch 22/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2668 - accuracy: 0.8824 - precision_5: 0.9670 - recall_5: 0.5958 - auc_5: 0.9469 - val_loss: 0.2907 - val_accuracy: 0.8694 - val_precision_5: 0.9745 - val_recall_5: 0.6782 - val_auc_5: 0.9483\n",
            "Epoch 23/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2629 - accuracy: 0.8846 - precision_5: 0.9688 - recall_5: 0.5930 - auc_5: 0.9481 - val_loss: 0.2257 - val_accuracy: 0.8964 - val_precision_5: 0.9744 - val_recall_5: 0.7581 - val_auc_5: 0.9695\n",
            "Epoch 24/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2613 - accuracy: 0.8839 - precision_5: 0.9706 - recall_5: 0.5931 - auc_5: 0.9480 - val_loss: 0.3006 - val_accuracy: 0.8564 - val_precision_5: 0.9614 - val_recall_5: 0.6984 - val_auc_5: 0.9460\n",
            "Epoch 25/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2573 - accuracy: 0.8854 - precision_5: 0.9715 - recall_5: 0.5920 - auc_5: 0.9492 - val_loss: 0.2328 - val_accuracy: 0.9037 - val_precision_5: 0.9805 - val_recall_5: 0.7310 - val_auc_5: 0.9670\n",
            "Epoch 26/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2581 - accuracy: 0.8863 - precision_5: 0.9718 - recall_5: 0.5904 - auc_5: 0.9490 - val_loss: 0.2967 - val_accuracy: 0.8620 - val_precision_5: 0.9686 - val_recall_5: 0.6823 - val_auc_5: 0.9468\n",
            "Epoch 27/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2560 - accuracy: 0.8880 - precision_5: 0.9731 - recall_5: 0.5917 - auc_5: 0.9497 - val_loss: 0.2924 - val_accuracy: 0.8628 - val_precision_5: 0.9605 - val_recall_5: 0.6995 - val_auc_5: 0.9485\n",
            "Epoch 28/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2560 - accuracy: 0.8873 - precision_5: 0.9741 - recall_5: 0.5901 - auc_5: 0.9494 - val_loss: 0.3532 - val_accuracy: 0.8325 - val_precision_5: 0.9307 - val_recall_5: 0.6899 - val_auc_5: 0.9317\n",
            "Epoch 29/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2516 - accuracy: 0.8893 - precision_5: 0.9749 - recall_5: 0.5920 - auc_5: 0.9506 - val_loss: 0.2329 - val_accuracy: 0.9045 - val_precision_5: 0.9840 - val_recall_5: 0.7320 - val_auc_5: 0.9671\n",
            "Epoch 30/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2455 - accuracy: 0.8938 - precision_5: 0.9769 - recall_5: 0.5875 - auc_5: 0.9528 - val_loss: 0.4247 - val_accuracy: 0.8016 - val_precision_5: 0.9495 - val_recall_5: 0.5279 - val_auc_5: 0.9002\n",
            "Epoch 31/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2429 - accuracy: 0.8946 - precision_5: 0.9787 - recall_5: 0.5849 - auc_5: 0.9534 - val_loss: 0.1830 - val_accuracy: 0.9291 - val_precision_5: 0.9875 - val_recall_5: 0.7742 - val_auc_5: 0.9805\n",
            "Epoch 32/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2454 - accuracy: 0.8926 - precision_5: 0.9778 - recall_5: 0.5843 - auc_5: 0.9522 - val_loss: 0.2939 - val_accuracy: 0.8621 - val_precision_5: 0.9784 - val_recall_5: 0.6683 - val_auc_5: 0.9473\n",
            "Epoch 33/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2391 - accuracy: 0.8966 - precision_5: 0.9801 - recall_5: 0.5856 - auc_5: 0.9543 - val_loss: 0.3060 - val_accuracy: 0.8479 - val_precision_5: 0.9654 - val_recall_5: 0.7036 - val_auc_5: 0.9452\n",
            "Epoch 34/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2409 - accuracy: 0.8960 - precision_5: 0.9793 - recall_5: 0.5846 - auc_5: 0.9535 - val_loss: 0.2921 - val_accuracy: 0.8612 - val_precision_5: 0.9681 - val_recall_5: 0.7054 - val_auc_5: 0.9491\n",
            "Epoch 35/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2411 - accuracy: 0.8946 - precision_5: 0.9799 - recall_5: 0.5849 - auc_5: 0.9532 - val_loss: 0.2386 - val_accuracy: 0.8912 - val_precision_5: 0.9828 - val_recall_5: 0.7337 - val_auc_5: 0.9651\n",
            "Epoch 36/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2401 - accuracy: 0.8953 - precision_5: 0.9803 - recall_5: 0.5823 - auc_5: 0.9535 - val_loss: 0.2681 - val_accuracy: 0.8772 - val_precision_5: 0.9849 - val_recall_5: 0.7070 - val_auc_5: 0.9567\n",
            "Epoch 37/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2387 - accuracy: 0.8957 - precision_5: 0.9811 - recall_5: 0.5850 - auc_5: 0.9540 - val_loss: 0.2949 - val_accuracy: 0.8572 - val_precision_5: 0.9830 - val_recall_5: 0.6890 - val_auc_5: 0.9512\n",
            "Epoch 38/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2405 - accuracy: 0.8943 - precision_5: 0.9809 - recall_5: 0.5866 - auc_5: 0.9534 - val_loss: 0.2454 - val_accuracy: 0.8930 - val_precision_5: 0.9838 - val_recall_5: 0.7111 - val_auc_5: 0.9626\n",
            "Epoch 39/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2347 - accuracy: 0.8988 - precision_5: 0.9823 - recall_5: 0.5810 - auc_5: 0.9552 - val_loss: 0.2730 - val_accuracy: 0.8799 - val_precision_5: 0.9762 - val_recall_5: 0.6876 - val_auc_5: 0.9538\n",
            "Epoch 40/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2302 - accuracy: 0.9009 - precision_5: 0.9846 - recall_5: 0.5798 - auc_5: 0.9567 - val_loss: 0.1903 - val_accuracy: 0.9311 - val_precision_5: 0.9906 - val_recall_5: 0.7604 - val_auc_5: 0.9777\n",
            "Epoch 41/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2342 - accuracy: 0.8990 - precision_5: 0.9824 - recall_5: 0.5828 - auc_5: 0.9552 - val_loss: 0.2553 - val_accuracy: 0.8869 - val_precision_5: 0.9793 - val_recall_5: 0.7268 - val_auc_5: 0.9617\n",
            "Epoch 42/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2305 - accuracy: 0.9015 - precision_5: 0.9837 - recall_5: 0.5806 - auc_5: 0.9566 - val_loss: 0.2925 - val_accuracy: 0.8659 - val_precision_5: 0.9738 - val_recall_5: 0.6981 - val_auc_5: 0.9500\n",
            "Epoch 43/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2313 - accuracy: 0.9000 - precision_5: 0.9837 - recall_5: 0.5818 - auc_5: 0.9564 - val_loss: 0.2071 - val_accuracy: 0.9217 - val_precision_5: 0.9899 - val_recall_5: 0.7272 - val_auc_5: 0.9735\n",
            "Epoch 44/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2272 - accuracy: 0.9019 - precision_5: 0.9850 - recall_5: 0.5779 - auc_5: 0.9572 - val_loss: 0.1653 - val_accuracy: 0.9431 - val_precision_5: 0.9931 - val_recall_5: 0.7657 - val_auc_5: 0.9843\n",
            "Epoch 45/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2266 - accuracy: 0.9028 - precision_5: 0.9853 - recall_5: 0.5758 - auc_5: 0.9574 - val_loss: 0.2722 - val_accuracy: 0.8800 - val_precision_5: 0.9846 - val_recall_5: 0.6925 - val_auc_5: 0.9561\n",
            "Epoch 46/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2247 - accuracy: 0.9030 - precision_5: 0.9854 - recall_5: 0.5779 - auc_5: 0.9579 - val_loss: 0.2992 - val_accuracy: 0.8679 - val_precision_5: 0.9747 - val_recall_5: 0.6899 - val_auc_5: 0.9487\n",
            "Epoch 47/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2249 - accuracy: 0.9033 - precision_5: 0.9858 - recall_5: 0.5765 - auc_5: 0.9573 - val_loss: 0.2739 - val_accuracy: 0.8748 - val_precision_5: 0.9778 - val_recall_5: 0.7201 - val_auc_5: 0.9569\n",
            "Epoch 48/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2210 - accuracy: 0.9066 - precision_5: 0.9864 - recall_5: 0.5708 - auc_5: 0.9584 - val_loss: 0.1823 - val_accuracy: 0.9261 - val_precision_5: 0.9906 - val_recall_5: 0.7564 - val_auc_5: 0.9803\n",
            "Epoch 49/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2237 - accuracy: 0.9046 - precision_5: 0.9867 - recall_5: 0.5708 - auc_5: 0.9574 - val_loss: 0.3208 - val_accuracy: 0.8345 - val_precision_5: 0.9761 - val_recall_5: 0.6975 - val_auc_5: 0.9425\n",
            "Epoch 50/50\n",
            "1542/1542 [==============================] - 6s 4ms/step - loss: 0.2164 - accuracy: 0.9098 - precision_5: 0.9872 - recall_5: 0.5667 - auc_5: 0.9596 - val_loss: 0.2669 - val_accuracy: 0.8785 - val_precision_5: 0.9792 - val_recall_5: 0.7089 - val_auc_5: 0.9588\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7fc5f0c33110>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 174
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 502
        },
        "id": "d99efa81",
        "outputId": "6511e1d9-2fed-404f-ff5c-0cac1b42ea4c"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "val_loss = model.history.history['val_loss']\n",
        "loss = model.history.history['loss']\n",
        "\n",
        "plt.figure()\n",
        "plt.plot(val_loss, label='teste_loss')\n",
        "plt.plot(loss, label = 'treino_loss')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "id": "d99efa81",
      "execution_count": 176,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAA10AAAHlCAYAAAAOfZZuAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdaXxb5Zk28OscSUe7ZHmRLTu2EyckMSELIQFCFqBAYUpCQmmbEqYbLR3gnWamb1sa2rcEuhJmWkpboC3DNoUurKUxS1pKC1nYQxayEztxEq+xbGtfz3k/yDJOvEmyZPmI6/+JykdHN8opP1+5n+d+BEVRFBAREREREVFOiPkugIiIiIiIqJAxdBEREREREeUQQxcREREREVEOMXQRERERERHlEEMXERERERFRDjF0ERERERER5ZA2lYuampqwbt069PT0oKioCBs2bMDkyZNPueaWW27BgQMH+v/3gQMHcO+99+KSSy7JasFERERERERqIqRyTtfnP/95XHPNNVi5ciWee+45PP300/jf//3fYa/fv38/vvCFL2Dz5s2QJCmrBRMREREREanJqMsLu7q6sHfvXixfvhwAsHz5cuzduxdut3vY9zz11FNYsWIFAxcREREREX3kjRq6WltbUV5eDo1GAwDQaDRwOp1obW0d8vpIJIKNGzfimmuuyW6lREREREREKpTSnq50vPzyy6isrER9fX3a7+3q8kGWR13tOG7Kyqzo7PTmuwwqYHzGKJf4fFEu8fmiXOLzRbmUi+dLFAWUlFiG//loN3C5XGhvb0c8HgcAxONxdHR0wOVyDXn9008/zS4XERERERFRn1FDV0lJCerr69HQ0AAAaGhoQH19PYqLiwdd29bWhnfffRcrVqzIfqVEREREREQqlNLywttvvx3r1q3DfffdB5vNhg0bNgAAbrjhBqxduxazZ88GADz77LO4+OKLYbfbc1cxEREREdFHUDweQ3d3J2KxSL5LUbWODhGyLGf8fq1WgsNRBo0m9Z1aKY2MHy/c00UfNXzGKJf4fFEu8fmiXOLzNbSTJ1thMJhgNtsgCEK+y1EtrVZELJZZ6FIUBX6/B6FQAKWlH263GvOeLiIiIiIiyr9YLMLAlWeCIMBstqXdbWToIiIiIiJSCQau/Mvkz4Chi4iIiIiIKIcYuoiIiIiIiHKIoYuIiIiIiNL24IO/QTQazdv7t29/B1/+8ucyfv94YugiIiIiIqK0PfzwA2MKTWN9v5qkPlyeiIiIiIgmhK27W7FlV2tO7r1kjguLZ7tGvOanP02c23vTTddDEETceefP8MgjD+Dw4UOIRCI4++wF+NrXvg6NRoOHHvotXn55EyRJD0EAfvGL3+C3v73vlPf/8pe/gSgK+OUv7x7yHql48cUG/OEPv4MgCKisnIRbbvkOHI5i7N69E3fffRdkWUEsFsP1138FH/vYx/Hcc8/giSd+D51OgqLI+P7370Rt7eQxfXfDYegiIiIiIqK0fOMb38azzz6J++9/CCaTCXfe+QPMmzcf69Z9D7Is4447/h+ef/4vuOiij+GJJ36P5557CXq9AYGAH5KkH/R+AMPe46qrrh61nsbGD/DrX/8KDz74GEpLS/HAA/fj7rv/C9///k/w+OOP4tprP4fLLrsCiqIgFAoAAO677x48/vjTKC0tRSQSGdOByaNh6CIiIiIiUpnFs0fvRo2nLVtew759e/DHPz4OAAiFQnA6y2E2W1BVVY0f/GA9zj33fFxwwVKYTOa07pGK7dvfwaJFi1FaWgoAWLnyk/jiF9cAAObPX4BHH30IJ04cx8KF52Pu3DmIxWTMn78QP/rReixevBSLFi1BVdWksX4Nw2LoIiIiIiKiMVLw4x//95DB5Te/eRi7d+/sG3zxr/jpT3+JadPOSOseY/GZz6zB4sXL8Pbbb+LnP78L5523CF/5yk348Y//C/v27cG7776DtWtvxDe/eSsWLVqc1c9O4iANIiIiIiJKm8lkht/vAwAsXrwMjz32KOLxOACgp6cHLS0nEAj40dPTg7PPPgdf/vK/oa5uKhobDw96/0j3SMX8+Qvw+utb0dV1EgCwceOfsXDhuQCA5uajqKqahFWrrsGnP30t9u7dg1gshpaWEzjzzLPwuc99Eeeeez4OHTqQnS9mCOx0ERERERFR2j772euwdu2N0OsN2LDhZ/jd7x7GF794LQRBgE4nYe3ab0Cr1eK7370FkUgYsixj+vSZuPDCiwe9/5e//A3+4z++gfvu+8Wge1RWVo1aS13dNNx447/j61//P32DNKrwrW99BwDw1FN/xPbt70Kn00Knk/DNb34bsizjRz+6HT6fF4Igory8HDfe+O85+64ERVGUnN09TV1dPsjyhCkHZWVWdHZ6813GuGnvDuCD470Tan1wofuoPWM0vvh8US7x+aJc4vM1tLa2o6ioqM13Gaqn1YqIxcY2NOP0PwtRFFBSYhn2ei4vpH6vvHsCDz6/D4FQLN+lEBEREREVDC4vpH5uTwgAcLzTh+nVRXmuhoiIiIgI+Pa3v4729vZTXisvL8eGDXfnqaL0MXRRP7c3DAA41sHQRUREREQTg5rC1XC4vJD6ub2JTtexDt8oVxIRERERUaoYuggAEIvL8PgiABi6iIiIiIiyiaGLAAA93jAUAGaDFic6J9YUSSIiIiIiNWPoIgAf7uc6q64EkZiM9u5AnisiIiIiIioMDF0E4MPJhXOnlgDgEkMiIiIiomxh6CIAH3a6Zk0phkYUcLyToYuIiIiIhvfgg79BNBpN+31btryKe++9JwcVAUuWLEAgMPFWbHFkPAEAuj1hGPVaWE0SKkpMONbO0EVEREQ0UUUPbkX0wGs5ubduxjLopi8e9bqHH34A1177Oeh0ulNej8Vi0GqHjxlLllyIJUsuHHOdasLQRQAS4+KLbXoAQLXTgoPHevJcERERERFNVD/96QYAwE03XQ9BEOFyuWC3F6G5+SgCgQAeeeT3ePHFBjzzzJOIx+OwWCz45jfXoaZmMl54YSO2bduMH/7wLmzf/g5+8Yuf4cwzZ2HPnt0ABNxxx48xefIUAMBjjz2CTZteAADU18/Cf/7nt2AymVKqcd++Pfj5z/8boVAQBoMR//mf30R9/Sy43W5873vfQXd3FwBgwYJzsXbtN7B7907cffddkGUFsVgMX/jC9bjssiuy8n0xdBEAwO0Jo9hqAABUl1nwxp52+IJRWIy6Ud5JRERERONNN31xSt2oXPnGN76NZ599Evff/xBMJhN+9KPbcejQQfzqV7+F0WjEzp3v4ZVX/oZ7730AkiTh9de34ic/+T7uv/+hQfdqajqM73znNtxyy3fx6KMP4tFHH8T69T/E669vxaZNL+DXv34IJpMZP/zhejzyyP/g5pvXjlpfNBrFd797C77znfVYsOBcvP32m/jud2/Bn/70Z2za9AKqqqpwzz33AQA8Hg8A4PHHH8W1134Ol112BRRFgc+XvZVfDF0EINHpmuyyAkh0ugDgeIcPM2sd+SyLiIiIiFTioosugdFoBABs3foaPvjgEL761S8CABRFgdfrGfJ9NTW1mD59JgBg1qzZ2Lp1MwDgnXfewiWXfBxmc+J306uu+iTuuee/U6qlufkodDodFiw4FwCwcOF50Ol0aG4+irPOmo0//OFx3HvvPZg3bz7OO28RAGD+/AV49NGHcOLEcSxceD5mzTorsy9iCAxdhGgsDm8gimLrh8sLgcQEQ4YuIiIiIkqFyWTs/2dFAa688ip85Ss3jvo+SdL3/7MoiojH4zmpL2n27Ll4+OHH8fbbb2LTphfw2GOP4P77H8RnPrMGixcvw9tvv4mf//wuLFx4Pr761Zuz8pmcXkj9kwuLbYnlhXaLHjaTjmPjiYiIiGhYJpMZfv/Qvy8uXrwUL730PDo62gEA8Xgc+/fvS+v+Cxaci1de+RsCAT8URUFDw5+xcOF5Kb23pqYW0WgU27e/AwB49923EYvFUFNTi5aWEzCbLbj00svxta99HQcO7Icsy2huPoqqqklYteoafPrT12Lfvj1p1TsSdroIbk8idDmsH/4tQ7XTwtBFRERERMP67Gevw9q1N0KvN8Dlcp3ys3nz5uOrX70Z69b9X8TjMmKxKC6++FLMnFmf8v0XLVqMw4cP4d/+7UsAgJkzz8QXvvDllN6r0+nwox/ddcogjR/+cAN0Oh22b38Hv//9YxBFDRRFxre+dStEUcRTT/0R27e/C51OC51Owte//q3Uv4xRCIqiKFm72xh1dfkgyxOmHJSVWdHZ6c13GTm3dXcrHnx+H3781fNRUZyYBvPEKx/g5XeP4/5vLINGZEM0Vz4qzxjlB58vyiU+X5RLfL6G1tZ2FBUVtfkuQ/W0WhGxmDyme5z+ZyGKAkpKLMNez9+mCd3eoTtdsbiMNncwX2URERERERUELi8kuL1hWIw66HWa/tc+HKbhRVWpOV+lERERERGd4uGHH8Crr/5j0Ot33/0rOBzFeahodAxdBLcn1D+5MKmixASNKOBYhw/nn5mnwoiIiIjoFIqiQBCEfJeRV1/60g340pduyNvnZ7I7i8sLKXEwct/kwiStRkRlqZnDNIiIiIgmCK1Wgt/vyeiXfsoORVHg93ug1UppvY+dLkK3N4Qzqu2DXq92WrDniDsPFRERERHR6RyOMnR3d8Ln68l3KaomiiJkOfNBGlqtBIejLL33ZPxpBc7tCeHFt4/h4+dUFfT0vnAkDn8oNmh5IZAIXdveb4MnEIHNlF6aJyIiIqLs0mi0KC11jX4hjSgf0zELN02MUVOrF0/+/RCa2wt7eZ3bGwIAFFsNg3724TCNwv4OiIiIiIhyiaFrGNXOxMS+5vbCPiMieTBysW3oThcAHCvw4ElERERElEsMXcMoLTLCqNeiucC7PMlOl8M2uNNlNUkoskg43lnY3wERERERUS4xdA1DFARMqbQVfJenu6/T5bAM7nQBwCSnhcsLiYiIiIjGgKFrBHVVdhzr8EEu4LGcbm8INrMEnXboR6HaaUHLST9i8cwnvBARERERfZQxdI2grtKOcDSOzu5gvkvJGbcnPOTkwqRqpwVxWUFrV2AcqyIiIiIiKhwMXSOYUpU4u+poAQ/TcHsHH4w8ULXTCgA41lG43wERERERUS4xdI2gptwKjSgU9J4mtycExwidropiI7QasaC/AyIiIiKiXGLoGoGk08BVYirYs7oCoRhCkfiQ4+KTNKKIqjIzQxcRERERUYYYukZRU25Fc4Eurese4WDkgar7JhgqBTxQhIiIiIgoVxi6RlHjtKDXF0GvP5LvUrLO7R3+YOSBqp0WeANReArwOyAiIiIiyjWGrlFUl/cNkijAYRpuT4qdrjILAHCJIRERERFRBhi6RlHtTASO5gIMHG5PGIIAFFmlEa+rLmfoIiIiIiLKFEPXKCxGHUpsBjQXYqfLG0KRRQ+NOPJjYDboUGzTM3QREREREWWAoSsFNeWWggwcox2MPFB1WWF+B0REREREucbQlYJqpwVtXQGEI/F8l5JVbm94xDO6Bqout6C1K4BorLC+AyIiIiKiXGPoGobc24aOP/8csr8bNeVWKACOdxZOp0dRFHR7Qii2jTxEI6naaYWsKGg5GchxZUREREREhYWhaziCCP+hdxDYeCdqrFEAhTVMwx+KIRKTU19e2D9QpPD2thERERER5RJD1zBEmxOua78HJdgLw6s/h8sQKqix8f3j4lPsdDmLjJC0Ivd1ERERERGliaFrBIZJM2C68ltQgl7cbH4J3e0t+S4pa5IHIztGORg5SRQFVJVZcJyhi4iIiIgoLQxdo9A4p8J05bdgFMJYFXwGsd7OfJeUFd0pHow8ULUzMcFQUZRclUVEREREVHAYulKgcdbhyKyvwIAI/H/5MWSP+oOX2xuGRhRgN498MPJA1U4L/KEYuvu6ZERERERENDqGrhSVTa3Hvd7LIEdDCGz8CWRPR75LGhO3J4QiiwRRFFJ+T3KYBvd1ERERERGljqErRa4SE9pQijcrrwNiEQQ23gm5tz3fZWXM7QnDkeIQjSSGLiIiIiKi9DF0pUirEVFZasaeHjOMy78NxKOJjldvW75Ly0i3N5zyuPgko16LUruBoYuIiIiIKA0MXWmocVrR3OGDWDwJxuW3AHI80fHqac13aWlRFAVubzjlcfEDJYdpEBERERFRahi60lBTboE3EEWPLwJNcTWMy9cBiozAxjsR71bPOHlvIIpYPPWDkQeqdlrQ3h1AOBrPQWVERERERIWHoSsNNeVWAMCxjsQhyZriqsRSQygINtyJePeJPFaXOrc3vYORB6p2WqAoQMtJf7bLIiIiIiIqSAxdaZhUlhgk0dz+4fI6jaMKxhXrAEFEcOOdiLuP56u8lLk9iZHvxSkejDwQh2kQEREREaWHoSsNJoMWZUUGNLd7T3ldU1QJ0/J1gKhBsGED4u5jeaowNe4MDkZOKi0yQi9pcKydoYuIiIiIKBUMXWlKDtM4nVhUAdOKdYBGi+DGDYh3NeehutS4vWFoNQIsJl3a7xUFAdVllv4lltmgKAp2N3YhGI5l7Z5ERERERBMFQ1eaasot6OgODhkQRHtFouOllRBo2IBYy748VDg6tycEh1UPUUj9YOSBqp0WHOv0Q1GUrNSz90g37n5iJ7a9r87x+0REREREI2HoSlN13zCN451DL68T7eUwrVgHwWBFsOEuhN96Coo8sTo4iTO60l9amFTttCAYjqGrNzTmWhRFwdOvHgbw4YAPIiIiIqJCwtCVphrn4GEapxNtTpg/eTt0M5YisqMBged+DNnTMV4ljsrtCWc0RCNpUhaHaWw/2IkjbYmlih5fZMz3IyIiIiKaaBi60uSw6mEx6gYN0zidoDPAcOH1MFx6M+TeNvifvg3Rg1vHqcrhybKCHl9mByMnTSozQ8DYQ5csK3jmtUa4SkyorbCi18/QRURERESFJ6XQ1dTUhNWrV+Pyyy/H6tWrceTIkSGve+GFF7BixQosX74cK1aswMmTJ7NZ64QgCAJqyi1DDtMYiq7uXJg/9QNoSmoQ+ucDCL7yGyiRQI6rHF6vP4K4rGR0MHKSQdKizGEcc+h6fU8bWrsCuHppHRwWPXrY6SIiIiKiApRS6Fq/fj3WrFmDTZs2Yc2aNbjtttsGXbN792786le/wkMPPYSGhgb8/ve/h9VqzXrBE0GN04oTnX7E4nJK14uWEhiXr4O04GrEDr8J/9PrEW//IMdVDi25b8oxhk4XkBymkXnoisZk/HlzE2orrDhnRhnsFgkef3hMNRERERERTUSjhq6uri7s3bsXy5cvBwAsX74ce/fuhdvtPuW6Rx55BNdffz3KysoAAFarFXp95t2Uiay63IJYXEabO/WOlSCK0M9fCdOKWwEoCPzlxwhv/wsUObXgli3dyYORx9DpAhKhq7M7iFAksyEhr+44gS5PCNdcWAdBEGA3S/AGooiP8/dBRERERJRr2tEuaG1tRXl5OTQaDQBAo9HA6XSitbUVxcXF/dcdPnwYkyZNwnXXXYdAIIDLLrsMN910E4Q0xpKXlFgy+FfIrbKywd26uTMVYONedAdiOHuIn498w/mQz/gZOl/6LfzvPAOxfR+cK/8DWntZlioeWXhvYqDHGVNKYTNLGd/nrGll+PPmJvijCqqr0vsOQuEYXnijGbOnluKihbUQBAGTKmxQAEhG/Zj2m6nRUM8YUbbw+aJc4vNFucTni3JpvJ+vUUNXquLxOA4cOICHH34YkUgEX/nKV1BZWYlVq1alfI+uLh9kOTtnP2VDWZkVnZ2DB2boBQU6rYg9H3Ridm1RRvcWLvgyDGX1CG39HZp/+39hWPYl6OoWjrXkUR1r7YWkFRHyhxAOZL6cz2ZIhPBdBztQYk7vkOWGbUfQ4wvj5qvPwsmTiSWKYt+ZX41H3YhXfHT+IzvcM0aUDXy+KJf4fFEu8fmiXMrF8yWKwogNpFGXF7pcLrS3tyMejwNIhKuOjg64XK5TrqusrMQVV1wBSZJgsVhwySWXYNeuXWMsf2LSiCImlZnHNEhCEATopi+G+ZN3QLSXI/TyvQi9+hCUaBg9vjCisXgWK/6Q2xuGw2ZIqwM5lBKbAUa9Nu3vwB+K4sU3mzFvWimmVdn7X7f3dd16ua+LiIiIiArMqKGrpKQE9fX1aGhoAAA0NDSgvr7+lKWFQGKv15YtW6AoCqLRKN544w3MnDkzN1VPANVOK5rbvVCUsXXmRHs5TCu/C2neckQPbIbv6dtw34PP4y9bj2Sn0NN0e0Jj3s8FJEJjdZkZxzrS+1uCF99oRigcw9XL6k553W7pC12cYEhEREREBSal6YW33347HnvsMVx++eV47LHHcMcddwAAbrjhBuzevRsAcOWVV6KkpASf+MQnsGrVKkybNg2f+tSncld5ntWUW+APxeD2jL0zI4ha6M/9FIzLb0EkGMRNhgZYD/8VSjyahUpP5faO7WDkgaqdVhzv8ENOMXj2+sJ4+Z1jOO/MclQ7T22/JjtdPTyri4iIiIgKTEp7uqZOnYonn3xy0OsPPPBA/z+Loohbb70Vt956a/aqm8BqnIl9R80dXpTYszP4QVtZj0fET2NB5GUsEt6E/6mjMCz5PLRVZ2bl/nFZThyMbM1OvdXlFoS3x9HZE0S5wzTq9Ru3HUFcVrBy6ZRBP9NpNTDptfCw00VEREREBSalThcNNslphgDgWPvYDgge6MRJP/a0RrGlaAV+7b0EsWgUwefvQvDvv4Yc6Bnz/Xu8ESgKstjpSnSrjqewr6uzJ4hXd7Rg6RzXsAHNbpG4p4uIiIiICg5DV4YMkhbOYhOaxzBM43Sbd7ZAIwq47rLp2BetwvYZ/w5p/krEmt6B/0+3IvL+38Z0rlf/wchZ6nRVlZohCEhpmMZzW5ogigJWLB7c5UqymyX0cnkhERERERUYhq4xqHFa0NyenXGTsbiMbe+3Yd4Zpagtt8JulnC4LQj9gqth/tQPoXHWIbztcQT+fAfiHY0ZfUZy/1m2Ol2SToOKYtOooetEpw+vv9+GS86ZBMcIQzzsFj0HaRARERFRwWHoGoOacgtO9oYQCI194MWOQyfhC0axdE4lBEFAXaUNjS29AACxqALGT3wThktughLoReDPP0Boy/9CCfvT+oxub1/oylKnC0gsMRwtdD27uQkGvQafOL92xOvY6SIiIiKiQsTQNQbVfcM0xnJeV9LmXa1wWPU4a0piFH9dpQ3t3UH4golAJwgCdFPPg/kzP4HurEsR3fcP+J+4FdGDW1MeW+/2hGCQNDAZsnYmNiaVJYNnbMifN7Z4sP1gJy4/twYW48iHKNstEsLROEKRoe9FRERERKRGDF1jUFueGCTRPMZhGm5PCO83dWHxbBdEMXFocZ3LBgBoavWccq0gGWG44DqYrr4dgrUUoX8+gGDDBsS7W0b/HG8YxbbsdbmAAcM0Oof+Dp557TCsJh0uW1A96r36D0jmEkMiIiIiKiAMXWNgt+hhM0toTvOA4NNt2d0KRQGWzHH1vzbZZYOARKdoKJrSWphW/j/ol34RcfcxBJ7+HsJvPQUlNvz0P3eWDkYeKBm6hur27Tvixt4j3bhy0WQY9aN31+zmRG1cYkhEREREhSR768w+ohLDNDLvdMmKgi27WlFf64CzyNj/ulGvRWWZedjQBQCCIEKqvwjayfMRfvNPiOxoQPSD12G44F+hqZ0HQRBOud7tDaOm3DLM3TLjsOphNmhx7LTgqSgKnnq1EcU2PS4+uzKle9ktfZ0uhi4iIiIiKiDsdI1RdbkFLSf9iMUzG+W+/2g3TvaGsHRAlyupzpUYpjHani3RaIPxohtgXHErBJ0Bwb/eg+BLP4Pc09Z/TTQmw+OPZHWIBpDYazbUMI0dh06iqdWDqxZPgU6rSeleyeWFPT6e1UVEREREhYOha4xqy62IywpaTqY3STBp865WmPRazJ9eNuhndZU2+EMxdHQHU7qX1jUDpmvugP78axFv+wD+p76L8JtPQIkE0d0XZEYa2Z6paqcVJzr9kOVEOJRlBc+81ojyYhMWz65I+T5mow4aUYCHnS4iIiIiKiAMXWOU3NOUyRJDfyiKdw904vxZ5ZB0g7tBdZV2AMPv6xqKIGohzbkc5tV3QjvtAkR2vgD/E7citH8rACXrgzSAxHcQiclo7w4AAN7c244TJ/24eukUaMTUHzFREGAzSxykQUREREQFhaFrjModJkg6MaNhGm/saUcsLmPpnKH3PFWVmqHXadIKXUmiyQ7jRV+GadX3IJgdcOx6DP9hfQmlckfa9xrNwGEasbiMZzc3oqbcggUznWnfi2d1EREREVGhYegaI1EUUF2W2TCNzTtbUFNuQW2Fddh7T3FZ0djam3F9GudUmFZ9DwerrkKZxgvTKxsQ2vwolNDYzxZLqiw1QRQEHOvwYfPOFpzsDeGaC6dCPG2QRyrsZgm93NNFRERERAWEoSsLqsutONbhTfmQYgA42uZFc4dv2C5X0pRKG5rbfYjG4hnXJwgi9upm4e7QNdDNuhTR/a/C96dvI7Ln71DkzAaADKTTauAqMeHwiV78ZdsRTJ9k7z/kOV12CztdRERERFRYGLqyoKbcgmA4jpO9oZTf89quFmg1Is6fVT7idXUuO+KyMuYDmLs9YRit9sTBytd8H5qSGoS3/g6BZ9cj1npgTPcGEksM9zf3oNcXwScvnDpoXH2q7GY9PIFI/1AOIiIiIiK1Y+jKghpnYnlgqsEoEo3jjT3tWDCjDGaDbsRr6yptAIDDGezrGsjtCaHYlphcqCmeBOOVt8Bw6c1QQn4EN/4EwVd+DdnfnfH9k/u65kwtwfTqoozvY7dIUBTAG2C3i4iIiIgKA0NXFlSVmSEIQHN7asM03j3YiWA4NuTZXKdzWPVwWPVobMl8XxeQOBh54ORCQRCgqzsX5tU/gTT/KsSa3oH/T+sQ3vFCRksO6yc7YDHqcM2FU8dUZ/KsLi4xJCIiIqJCoc13AYVAr9Ogotg06IDg4Wze2YJSuwEzah0pXV9XactogmFSOBqHLxgd8owuQauHfsEnoZu+BOHX/4DIW08gfmwXDB/7NzCj+rcAACAASURBVIjm1OoDgMkVNtyzdknGywqT7OZEjQxdRERERFQo2OnKktpya0pj4zu6A9jf3IOlc1wpT/erq7ThZG8IngyX3HV7E9MAi0c4GFm0OWH4+FoYLvoK4p1NCDz1PcSOvpfW54w1cAGJ5YUAeFYXERERERUMhq4sqS63wO0JwxeMjnjdlt2tEARg8ezRlxYmTc3gkOSBuj2JAR+jHYwsCAJ005fA/MnbIViKEdx0D0LbHocSH/nfKZts/csLOTaeiIiIiAoDQ1eWJIdpHBthX5csK9i6uw2z60pGDUAD1VZYIQpCxqHLnex02YbvdA0kFrlgWvU96M66DNH3/4bAn38Auac1o89Ol16ngVGvYaeLiIiIiAoGQ1eWVJcnpvcdHWGC4ftNXej2hlMaoDGQXqfBpDJzxsM03MlO1wjLC08naHQwXHAdjJf/BxSfG/5nbkf04Ja0ziLLlN2s554uIiIiIioYDF1ZYjNJcFj1ODbCvq7NO1thNekwd1pp2vevq7ShqdUDOYPQ4/aGYTXpoNNq0n6vtvZsmD71A2jKJiP0z/9B6B+/hRIJpn2fdNjNEnp9XF5IRERERIWBoSuLqp0WNA8zwdDjj2DHBydxwVkV0GrS/9qnVNoQDMfR1hVI+71uTxjF1tSXM55ONDtgvPLbkBZcjdjhN+B/Zj3inU0Z3280dovEThcRERERFQyGriyqKbeg9WQA0Vh80M+2vd+GuKxg6ZzKjO5dN4ZhGm5vaMhx8ekQRBH6+SthXHErEI8h8NwPEdn1IhQl/TO9RmMzM3QRERERUeFg6MqiGqcVsqLgeKf/lNcVRcHmXS2YWmVDZak5o3u7Skww6jVobM0gdHnCKQ/RGI22YjrM13wf2pq5CL/xJwRfuhtyMPMzxIZSZNEjFIkjHBkcXomIiIiI1IahK4uSwzROPyT5cIsHrV2BjLtcACAKAqa4bGkP0wiGYwiGY2lNSxyNYLDAcNnXoF/8OcRb9iXO9DqxN2v3t3NsPBEREREVEIauLCorMsIgadB82tj4zTtboNdpsHCmc0z3r6u04XiHH+Fo6h2gVA5GzoQgCJBmXQLTqvUQ9GYEn/8vhN96KivLDT8MXVxiSERERETqx9CVRaIgDBqmEQzH8Na+Diysd8Ko147p/nUuO2RFwdG24Sckns7tTe1g5ExpSqphuno9dDOWIrKjAdHdm8Z8T7slERB5VhcRERERFQKGriyrcVpxrMPXP9r97f0dCEfjWDaGpYVJdZU2AOkN03B7ctPpGkjQ6aFf9iVoa89G+O2nEXefGNP92OkiIiIiokLC0JVl1eUWhCNxdHYnzrLavKsFrhITplbZxnxvm1lCqd2Q1r4utycEAUBRDkMXkFhuqF/6RQg6I0L//C0UOZbxvSxGHURB4J4uIiIiIioIDF1ZVltuBQA0d/jQctKPwyc8WDqnEoIgZOX+dZW2tCYYur1h2CxSRmeDpUs02aFf8nnIJ48isn1j5vcRBVjNOi4vJCIiIqKCwNCVZZWlJmhEAc3tXmze1QKNKOCCsyqydv+6SjvcnjB6fKl1gbo9oZwuLTydrm4htNMWIfLexjEdoFxk1nN5IREREREVBIauLNNpNXCVmNDU6sG299swd1opbH17lLIh3X1dbm8YxdbcDNEYjmHxv0Iw2RH6xwNQYpkFJ7tFYqeLiIiIiAoCQ1cOVDut2HukG95AFEvnuLJ679pyCzSigMMp7OtSFAVuTxiOLB2MnCpBb4Zh2fWQe1oQfvvpjO5hM0vc00VEREREBYGhKwdq+g5JLrJIOKuuOKv31mk1qHZa0JRCpysYjiEcjY97pwsAtNWzoau/GNHdf0WsZX/a7y+ySPD4o/1TIImIiIiI1IqhKwdq+oZpLJ7tgkbM/ldcV2lDU5sXsjxyIOkfFz/Ona4k/fmrIVhLEXr1f6BEgmm9127WQ1YU+ALRHFVHRERERDQ+GLpy4IxJdqxaMgUfX1idk/tPrbQjHImj5aR/xOtyfTDyaASdAYaLb4Di7UL4jT+m9V6e1UVEREREhYKhKwe0GhFXLZkCqyl7AzQG6h+mMcro+PE4GHk02orp0M25AtH9ryLWvDPl99n6Qxf3dRERERGRujF0qZDTYYTZoB31kGS3NwRREFBkyV/oAgD9gqshOqoQeu1hKCFfSu8psvSFLk4wJCIiIiKVY+hSIUEQMKXShsOjDNNwe8KwWySIYnYOZs6UoJVguPirUIJehLb+LqX32Li8kIiIiIgKBEOXStW5bGjp9CMYjg17Tbc3nLchGqfTlNZCOucqxA6/iejht0a93iBpoZc07HQRERERkeoxdKlUXaUdCoAjbd5hr3F7QnkZFz8cad5yiGVTENryKORAz6jXF/GsLiIiIiIqAAxdKtU/TGOYfV2KosA9gTpdACCIGhguvgGIRRL7u0Y5g8tultjpIiIiIiLVY+hSKYtRh3KHEY3D7OvyBaOIxuQJ1ekCAE1RJfTnfgrx5p2IHnhtxGttFj33dBERERGR6jF0qVhdpQ2NLZ4hO0b5Phh5JLqzLoPGNRPh1/8A2ds57HV2s8TQRURERESqx9ClYnWVdvT6I/0Ba6B8H4w8EkEQYbjoywCA0D//B4oiD3ldkUVCMBxDJBofz/KIiIiIiLKKoUvFRjokORnEHHk8GHkkorUM+kXXIt56ANH3/zbkNRwbT0RERESFgKFLxaqdFmg14pDDNNzeEDSi0B9cJiLdjGXQ1MxF+K2nEO9pGfRzuzkRGBm6iIiIiEjNGLpUTKsRUVthGXKYRrc3DIdVD1HI78HIIxEEAYZlXwK0EkL/eACKfOoywiJLX6eLEwyJiIiISMUYulSuzmXH0TYvYvFT90W5PWEUT9ClhQOJpiIYlnwBcmcTontePuVn9v7lhTyri4iIiIjUi6FL5eoqbYjEZJzo9J/yutsTmpBDNIaim3ouNJX1iOx4AUrsw66W1SRBENjpIiIiIiJ1Y+hSuaEOSZYVJbG8cAKOix+OdM4qKMFeRPf9o/81URRgNXFsPBERERGpG0OXypXaDbCadKfs6/L6I4jLyoQ7GHkkWteMIbtdRWYJvT4uLyQiIiIi9WLoUjlBEFDnsp0yNt7tnbgHI49Emr+yr9v1z/7XbBZ2uoiIiIhI3Ri6CkBdlR2tXQEEQlEAif1cAFTV6QIAbeVMaFwzEdn5YbfLbmboIiIiIiJ1Y+gqAMl9XU2tXgADDkZWWacLAKRzVkIJ9CC6/1UAQJFFD48/AllR8lwZEREREVFmGLoKwJQKGwR8OEyj2xuGTivCatTlt7AMaCvroXHNQGTH81BiEdjMEuKyAn8wmu/SiIiIiIgywtBVAEwGLSpKTP3DNNzeEBxWPYQJfDDySKRzVvV3uz48q4tLDImIiIhInRi6CkRdpQ2HWzxQFEU1ByMPR+Oa2d/tshsSwZGhi4iIiIjUiqGrQNRV2uELRtHZG4Lbq56DkYciCEJikmGgB6Wd7wAAx8YTERERkWoxdBWIOldimMbh473o8UZUNy7+dJrKemgqpkM6+DdoEWeni4iIiIhUi6GrQExymiFpRbx3qBOyoq6DkYciCAKkc1YBgW4sMX2AXh9DFxERERGpE0NXgdCIIiZXWLGrsQsA4FDxnq6kZLfrEv378PkC+S6HiIiIiCgjDF0FpK7SjkhUBgBV7+lKSu7tsgl+uHrey3c5REREREQZYegqIMlDkgGofk9XkqbqTHRoKzEv/DaUOM/qIiIiIiL1YegqIMnQpddpYNJr81xNdgiCgMMlF8IGP6IHNue7HCIiIiKitDF0FRCHVQ+7RUKxTb0HIw8lXDodjdEyhN9rYLeLiIiIiFSHoauACIKAC+dW4pwZznyXklV2ix4vBecCfje7XURERESkOoWxBo36rVpal+8Sss5ulnAg5kLUMRnCew3QzVgGQcNHl4iIiIjUgZ0umvCKLHoAAjqqL4XCbhcRERERqUxKoaupqQmrV6/G5ZdfjtWrV+PIkSODrvnlL3+JRYsWYeXKlVi5ciXuuOOObNdKH1E2swQAaJUmQ3RORWRHA5R4LM9VERERERGlJqXQtX79eqxZswabNm3CmjVrcNtttw153apVq/Dcc8/hueeew/r167NaKH102cw6CAB6/RHoz1kJxdeF6MEt+S6LiIiIiCglo4aurq4u7N27F8uXLwcALF++HHv37oXb7c55cUQAoBFFWE06ePwRaCbNhuisQ+S9jex2EREREZEqjDqNoLW1FeXl5dBoNAAAjUYDp9OJ1tZWFBcXn3Lt888/jy1btqCsrAxf+9rXcPbZZ6dVTEmJJa3rx0NZmTXfJRCAYrsRwagMp9OGwMXXou1PP4Kh9R3Yzr4s36WNGZ8xyiU+X5RLfL4ol/h8US6N9/OVtRFwn/3sZ3HjjTdCp9Nh69atuPnmm/HCCy/A4XCkfI+uLh9kWclWSWNWVmZFZ6c332UQALNBiw53AJ2dXii2aRDL6tC1+SmEXAtUPcmQzxjlEp8vyiU+X5RLfL4ol3LxfImiMGIDadTlhS6XC+3t7YjH4wCAeDyOjo4OuFyuU64rKyuDTqcDACxevBgulwuHDh0aS+1E/exmCR5/GEDiPDL9OSuheE8iemhrnisjIiIiIhrZqKGrpKQE9fX1aGhoAAA0NDSgvr5+0NLC9vb2/n/et28fTpw4gSlTpmS5XPqoslsk9PojUJREJ1RTPQdi2ZTE3i6Ze7uIiIiIaOJKaV3W7bffjnXr1uG+++6DzWbDhg0bAAA33HAD1q5di9mzZ+NnP/sZ9uzZA1EUodPpcNddd6GsrCynxdNHh92sRyyuwB+KwWLU9Xe7gi/9HNGDWyHNvDDfJRIRERERDSml0DV16lQ8+eSTg15/4IEH+v85GcSIcsHed1ZXrz8CizGxjFVTPbev29UA3fTFEET17u0iIiIiosKV0jldRPlWZEmELo8v3P+aIAjQz18JxduJ2MFt+SqNiIiIiGhEDF2kCra+TlePP3LK65qauRBLJyO8owGKHM9HaUREREREI2LoIlWwm/UAgF7fqaFLEARIZy+H4ulArPHtfJRGRERERDQihi5SBaNeA51WhOe0ThcAaCfPh+ioROS9BiiKnIfqiIiIiIiGx9BFqiAIAuxmCT3+8BA/EyHNWw65+zjiR3fmoToiIiIiouExdJFq2C3SoOWFSdqp50GwliH83sb+s7yIiIiIiCYChi5SDbtZP+TyQgAQRA2kuZ+A3NmIeMu+ca6MiIiIiGh4DF2kGnaLhN5hQheAxFldpiJE3ts4jlUREREREY2MoYtUw26W4AtGEYsPPSxD0EqQ5lyBeMs+xNs/GOfqiIiIiIiGxtBFqmHvO6truCWGAKCrvwiC3oIwu11ERERENEEwdJFq9J/VNULoEnQG6GZfhnjzTsS7mserNCIiIiKiYTF0kWrYLYlOV49v8Nj4gaRZlwI6AyLvNYxHWUREREREI2LoItVILi8cqdMFAILeDGnWJYg1vg25p208SiMiIiIiGhZDF6mGLbmna5izugbSnfVxQKNFeMfzuS6LiIiIiGhEDF2kGlqNCItRN2qnCwBEkx26mRcidmgbZF/XOFRHRERERDQ0hi5SFbtFGnVPV5I0918AAJGdL+SyJCIiIiKiETF0karYzdKII+MHEi0l0E2/ANH9r0EO9Oa4MiIiIiKioTF0karYzVJKywuTpHlXAnIM0d2bclgVEREREdHwGLpIVewWPXp8ESiKktL1or0C2ikLEdn7CpSwP8fVERERERENxtBFqmI3S4jFZQTDsZTfI529AoiGENnzcg4rIyIiIiIaGkMXqUqqZ3UNpCmphqZmHiK7/wolGspVaZRDsqLg2dca0e1NbYgKERER0UTC0EWqYrfoAQC9KZzVNZD+7OVA2I/ovn/koizKsbauADZuO4J/vnci36UQERERpY2hi1Ql2enq8afX8dCUT4Omsh6RXZugxNILbJR/vX3HBOxr7s5zJURERETpY+giVbFbEqHLk2anC0js7VICPYge3JLtsijHkstJm1o8CEVS389HRERENBEwdJGqmPRaaDViWnu6kjSV9RCddYjsfAGKzF/c1aSnL2THZQWHjvPMNSIiIlIXhi5SFUEQYDdL/b+Ep/te/dkroHhPIvbBmzmojnLF449AqxGgEQXsO8olhkRERKQuDF2kOnaLBE+ae7qSNDVzIRZPQmTH81AUOcuVUa70+MMosugxtcqO/QxdREREpDIMXaQ6drOU0fJCABAEEdK85ZB7WhA7sj3LlVGu9PoisFsk1Nc6cLTdC38omu+SiIiIiFLG0EWqY7foM1pemKStOxeCrRyR9zZCUZQsVka50uuPoMisR32tA4oCHGzuyXdJRERERClj6CLVsZsl+IJRxOKZLQ8URBHSvE9APnkU8eO7s1wd5UKvLwybRUJdpQ2SVuS+LiIiIlIVhi5SneRZXd5A5kvMdGcshmAuRuS9hmyVRTkSjcnwh2IoMkvQakScUV3E87qIiIhIVRi6SHWSoas3w2EaACBotJDm/gvibQcRaz2QrdIoB5J/znaLHgBQX+vAiU5/xvv6iIiIiMYbQxepTvKX77Hs6wIA3cxlEAxWRN7bmI2yKEeS4SoZtutrHQCAA+x2ERERkUowdJHqJH/59oyx0yFo9dDNuRzx4+8j3tGYjdIoB3r7wnVRX9iuKbfAqNdyXxcRERGpBkMXqY4tubzQl/nywiTpzEsAvRnh7X8Z870oN5J/zsk/d40oYkZ1EUMXERERqQZDF6mOTivCbNCiJwt7egTJCGn2xxFv3oH4yaNZqI6yrdcfgQDAZtb1v1Zf60BHdxBdvaH8FUZERESUIoYuUiW7RQ/PGPd0JUmzLgUkIyJpdrticRlNrR7IPOsrp3p8EVhNOmjED/9zldzXtZ/7uoiIiEgFGLpIlexmKWvT6wS9GdJZH0fsyLuIu4+l/L5/bD+BHzz6DtY/+Ba2vd+a8blhNDKPP9I/PCWpsswMq0nHJYZERESkCgxdpEqJ0DX2PV1J0lmXAToDIttTn2R48HgPrCYdBAH4n4Z9uPU3b+Dv7x5HOBrPWl0E9PjC/cNTkkRBwMwaB/Yd7YbCTiMRERFNcAxdpEp2i4ReXyRrv3ALBgukWZci1vg24t0tKb2nqdWD+loH7rj+XKz91Bw4rHo8/reDuOX+bdi47QgCocwPb6YP9fojsFukQa/X1zrQ7Q2jozuYh6qIiIiIUsfQRapkN+sRickIRbLXVdLN/jig1aV0blePLwy3J4y6SjsEQcC8aaX4zufOwbrr5mOKy4ZnX2vEN+/bhif+8QF6sjBl8aNKVpTE8kKzftDPZvbt6+ISQyIiIprotPkugCgTyeVmvf4IjPrsPMai0QbdmR9DdPcmyOeshGivGPbaxhYPAKDOZTvl9enVRZheXYTmdi9efLMZm95qxsvvHMPi2S78y3k1cDpMWan1o8IXjCIuK0N2usodRjiseuw72o2Lzq7KQ3VEREREqWGni1Qp+Ut4Ns7qGkiacwUgahF+r2HE65paPdCIAmrKLUP+vKbcin+7ahZ+8tXzsWS2C1t3t+LW376BXz/3PprbvVmtuZB5TjsYeSChb1/X/uZuTpAkIiKiCY2hi1RpYKcrm0RTEXT1FyF2aBtkT+ew1zW2eDCpzAJJpxnxfk6HCZ+/YibuuukCXHFuDXYd7sLtD7+Nu5/YiYPHerJaeyHq6RuWcvogjaT6Wge8gShaOv3jWRYRERFRWhi6SJWSI8R7s3RW10DS3E8AoojIjqG7XbKioKnVg7pK25A/H0qRRY9PXzwN/33zBfjksjocafPgzse3Y/v+jmyVXZCSf75DLS8EgJm1RQC4r4uIiIgmNoYuUiWzQQuNKGS90wUAotkB3YxliB7cAtnXNejnrV0BhCLxtEJXksmgw/ILJuOumy6A3SKhYWtjNkouWMk/3+E6XaV2I5xFRoYuIiIimtAYukiVBEHoGxufm8mA0rwrAQCRHc8P+llT3xCNKa70Q1eSXqfBktkuvLuvHW5PKOP7FLpeXwR6SQODNPywlJm1Dhw41o24zMOpiYiIaGJi6CLVShyQnP1OFwCIlhLopi9BdP9rkP2ndlEaWz0w6jWoKBnbJMKlcyshK8CWXa1juk8h6/UPPhj5dPW1DgTDcTS3+8apKiIiIqL0MHSRatnN+pyFLgCQ5i0HFBmRnS+c8npTiweTK2wQBWFM93cWGTHvjDJs3tUCWeb0vaH0+CIoGiV08bwuIiIimugYuki1crm8EABEWxm0Z1yA6L5/Qg4kJg1GonEc7/RltJ9rKJcvqkWXJ4w9R9xZuV+h6fVH+oemDMdullBVamboIiIiogmLoYtUy26W4A1Ec7qXR3/2ckCOIbLrJQBAc7sPcVkZdChyps6b5YLVpMOrO1qycr9C0+sbfXkhkOh2HTrWg1ic+7qIiIho4mHoItWymyUoALyBaM4+Q7RXQDv1fET3vgI56EFjSy8AYEqWOl06rYjFZ7mw84OTOe3aqVE4EkcoEh92XPxA9bUORGIyGvuGnBARERFNJAxdpFo2c+7O6hpIOnsFEIsiuvuvaGz1oNimR9EoS97SsWxeJeKygi27OVBjoN7+g5FH/65n1BRBAPd1ERER0cTE0EWqVdTXAUn+cp4rGkcltHULEdnzMlpaOrK2tDCpotiEGdVFeG1nC2SFAzWSkkNSilLodJkNOtRUWLGfoYsoZV29IS7JJSIaJwxdpFrJvT657nQBgDR/BRANYXZkR9aWFg60bF4lOntCDA0DJP9cbSns6QISSwwPt/QiHI3nsiyighAMx/CdB97A6++35bsUIqKPBIYuUi17f6cr96FLU1wNf9lsLNPvw9RSXdbvv2BGGcwGLV7byYEaST19e9xSXcpZX+tALK7ggxO9uSyLqCC4PSFEYzI6e4P5LoWI6COBoYtUS6fVwKTXjkunCwDet5wPkxhFlfvNrN9bp9Vg0VkV2H6wE57A+Pz7THS9/ghEQYDFlFrIPWOSHRpRYLeQKAU9ff/d9OVwEBEREX2IoYtUzW6Rcr6nK2mX24wPMBny3pehRLL/t8MXzq1ELK5g224u9wESywttZl3Kh1AbJC2mVNom1DCN9w524vaH3+K+GZpwkp1kb5Chi4hoPDB0karZzdK4LC9UFAVNrR40O5cBYT8ie/+e9c+oKrNgapUNr+1sgcKBGikdjHy6+hoHmlo9CIRiOaoqPXuPdKO53YeTvaF8l0J0iuR/N9npIiIaHwxdpGo2szQuyws7uoPwh2JwTJkJTfVsRHdtghLNfoftwrlVaHMHcOg49yWlejDyQPW1DigKcPB4T46qSk9yv0y7O5DnSohO1eNlp4uIaDwxdJGqFVn049LpamxNHLo7xWWDfv5KKCEvovteyfrnLJzphFGvwas7TmT93mrT64+kNC5+oKlVNui04oTZ19XR3Re6ujmsgCaWnv5OF/eQEhGNB4YuUjW7WUI4GkcoktvlZI0tHuh1GlSVmqEpnwZN1ZmI7HwRSiy7v7DoJQ3OP7MC7xzohD/00f0baFlW4AlE+g/ATpVOq8G0KvuE2NclywpOJjtd3ex00cSS3NPlC8Z4PiAR0Thg6CJVS57hlOtuV1OrB7UVVohiYqiDNH8llKAH0f2vZv2zLpxXiWhM/kifn+MJRKAoqR2MfLr6WgeOdfjgzfPf4Pf4wojFE7/MdnB5IU0wvX2hS1YUBMMTYw8kEVEhY+giVUue4ZTLfV3RmIzmdi/qBhyKrHXNgMY1A5Edz2e921VTbsXkCitezdFAjeZ2L1pO+rN+32xK/nna0+x0AYnQBQAHmvO7ryu5tNBmlri8kCYURVH6poMm/lKDwzSIiHKPoYtUzT4Ona7jnT7E4grqXLZTXpcWfBJKoAeRXS9m/TOXzavEiU4/Gls8Wb3v8Q4ffvLYdjywcW9W75ttyWMA7Bl0umorrNBLmrwvMezoSQStWZOL0dV3EC3RRBAMxxCJyaguMwPgMA0iovHA0EWqZuv7pTy5VCYXksFnYKcLSHS7tFMWIPLe85B9XVn9zPPqy6HXafDqzpas3dMXjOIXT+9COBrH0XZv/56OiSjZ6SpKc3ohAGg1ImZUF+U9dHX2BKERhf6Jisn9XUT5ljwYeZLTAgB5X4pLRPRRwNBFqmYx6qARhZx2uhpbPLCbJTisg5e66c9fDUBG+K0ns/qZRr0W553pxFv72rOy3yIuy7j/z++jxxfG5y6fAQDY3ZjdoJhNyclqmXS6AGBmjQNt7gC6vfkLlp09QZTYDHCVmAAA7W6GLpoYkn/hMqksEbq4vJCIKPcYukjVREGAzSyhK4eHzza2elBXaYMgCIM/31oGac6/IPbBG4i1Hcrq5y6bW4VIVMYbe9vHfK8nXjmMfUe78bnLZ+CieZWwWyTsbnRnocrc8PgiMOm10Gk1Gb0/ua8rn6PjO7qDKHMYUV7cF7o4wZAmiGQnuT90cXkhEVHOMXSR6s2scWDn4S5EY/Gs39sfiqLdHcCU0/ZzDSTNWw7B7EB42+NQlOzt25nismJSmQWv7RjbEsMtu1rxt3eO4dIFk7B0TiUEQcDsuhLsaXIjLk/MfUY9/nDGXS4AqC63wGzQ5nWJYWdPEGVFRliMOpgNWg7ToAmjp2/PpNNhhE4rck8XEdE4YOgi1btgdgWC4RjeO3Qy6/duah16P9dAgk4P/bmfhnzyCGIHt2btswVBwIXzKnG03YsjbZkN1Djc0ov/3bQf9bUOrP7YtP7X59SVIBiO4fCJ7A7qyJZeX6R/SEomREHAzBoH9h1152QC5GgCoSj8oRicRUYAgNNhQjvHxtME0eONQK/TwKjXwmLUcXkhEdE4YOgi1auvccBh1WNbDs61amrxQAAwuWL40AUA2mmLIJZPQ/itJ6FEstfRWDSrHJJWTMRErAAAIABJREFUzKjb1e0N41fP7EaRRY+bVp0Fjfjh/93PnFwMjShg1+GJua+r1x/uPw4gUzNrHejyhNGZw6Wnw+nsSXxmWV/oKi82ooPLC2mCSPz/K/GXGlajjssLiYjGQUqhq6mpCatXr8bll1+O1atX48iRI8Ne29jYiLlz52LDhg3ZqpFoRKIoYNGsCrzf6M76FMPGFg8qSkwwGbQjXicIAgwXXAcl6EF4+1+y9vkmgw4LZjrxxt52hCKpD9SIxuL41TO7EQrHsfaaObAYdafdV4tpVfYJOUzj9DOEMpXPfV3JcfFlRQYAQLnDBLcnnJMlsETp6vGGYe/7Sw2rScfphURE4yCl0LV+/XqsWbMGmzZtwpo1a3DbbbcNeV08Hsf69etx6aWXZrVIotEsnl0BWVGyMnQi6f+zd+fhbdVnvsC/5xztiyVZsmV5XxI7duzsQAiEQIDSlgSYcls6dLtlCnSm09k6S5fpdLszbWfuLLft0JlCB0opbWEoBTKUNQGyh5DVcbzGieNVsq19X879Qz4iiTct50iK/X6ep89DE/no2JHk85739/u+PM9jcMwzaz7XfLiyBsiatyLa+RoSbvG6btvWVSIUiePds/a0Hs/zPJ58pQeDYx58bkdbKhb6Sh1NZly0+wqa8DeXUCSOSCyRc6fLZtbAoFUUZF+X0NVKdbpMavB4f2AyIYXk8kdSnS6dRkF7ugghJA8WLbqmpqbQ1dWFHTt2AAB27NiBrq4uTE/PTj77yU9+gptvvhn19fWinyghC7GZtWiwlWD/afGKnSl3CJ5AdMH9XFdSXnsvwMkROvgr0c5jRZUBNrMm7Zldrx8dxv7Ocdx9YwM2tpTN+7g1jWYAxRcdL8RZ57KnC0h2H1vrTDh7wZn3fV0OVwglGjnUymSH9P0EQyq6SGEJnWThpgbt6SKEkPxYeM0UgLGxMVitVnBcMrqZ4ziUl5djbGwMpaWlqcd1d3dj3759ePLJJ/HII49kdTJm89x35AuprExf6FMgabpjcx3+4/nT8EUTaKg05Hy87pmQiQ1ttgxeB3q4tn4U07t/Dq2nH5qm9Yt+RTrH/vANjfjpi53wx3jUL9B5O9FrxzO7+3B9hw0P3N0Blp0dcy+wWHQwG1ToGXbj3ttaFj2HfBn3JIuuuipjzu+/a9ptyaWZCaC2In/vZZc/AluZLnX+al1ymaE/Es/7Zwp9hpFLBUJRhKNxVFn1KCvTw2rRIRCOwVSqhYzLfJs3vb6IlOj1RaSU79fXokVXOqLRKL7+9a/ju9/9bqo4y8bUlA+JRP6TxuZTVqaHw+Et9GmQNLXVGsGxDHa9M4CP37oy5+Od6J6AjGOhlTMZvQ74hpvAlLwK+yv/Bc3/+jYYdv63WbqvsbUNJsg4Br/d04dP3N4852PszgC+97OjsFm0+ORtKzE15Vv0uKvrS3Hk7ATGxt1ZXXBJ4cKwCwDAx2I5v/+qS5PL+w6cGIF64/wFqNhG7D6srDFcdv46tRznhl15/UyhzzBypbEpPwBAxgAOhxfszJiL8xedGXeX6fVFpESvLyIlKV5fLMss2EBa9CrLZrNhYmIC8XhyA3g8HofdbofNZks9xuFwYGhoCA899BC2b9+On/3sZ3jmmWfw9a9/XYRvgZD06NRyrFthwaEz46LMnxoc86CuQpdxMcJwcqg2/z4SrlFEu/bkfB5A8nvb2FKOg53jiERnhzEEwzH88LnTAIAv3rsmtaxtMR2NZoQicfQPu0U5TzEIYSiGHPd0Ack9VRaDKq/7umLxBKa9oVRcvMBaqqbYeFJwrpnByMaZAksI2fFRmAYhhEhq0atJs9mM1tZW7Nq1CwCwa9cutLa2Xra0sLKyEocPH8bu3buxe/dufOYzn8HHPvYxfOc735HuzAmZw5b2CngCUXSem73nMBOxeAIXxr1otGW3TJGrWweuuh3ho88jERLnTspNaysRCMdwtOfyQI0Ez+OxXV0YmwrgD+9pn3Wxv5C2ehM4limqfV1ufwQyjoF2kcTIdK2qM6FnyJm3LvqUOwSefz9EQ1Bu1NCeLlJwwk0No15IL0wWX17a10UIIZJK6xb+N7/5TTz11FO444478NRTT+Fb3/oWAODBBx/E6dOnJT1BQjLR0WSGTi3H/hxndo04/IjEEmiozG69L8MwUF7/+0A0hMi7v8npXASrao0oN6lnzex6cd8gjvdN4r7tK9BWXzrPV89NrZRhZbUBp4qo6HLNDEZmGHGWA7bWmeAPxXDRvvhySzG8Hxc/u9Pl9IYRnqNTSUi+CJ0ug3am6BI6XZRgSAghkkrrVnJTUxOeffbZWX/+6KOPzvn4L37xi7mdFSFZknEsrmuz4u0To/CHotCq5It/0RwGx5IhGo05BHJwpirIV9+K6Jk3IG+7BZy5NutjAclCbtvaSjz71gDGpvywmbV4r8eOF/efxw0dFbhtU3VWx13TZMEze/ox7QmhtESV0zmKweMPi7K0ULCqNjmv6+wFJ+ryEKbhmCm6yk1XFF2mZIKhwxmcN8afEKm5fGEoZCzUyuT+a50m+RlJsfGEECKt4tg5T4iIbuioQCyeSHuu1VzOjXqgU8tRZsitCFFuvAeMQovwgadFiS3f0mEDxzJ4+8Qohu0+PLbrLBorS/DpO1qy7gx1NCa7Y8XS7XL5IznHxV/KpFei3KjGudH87FuzO4NQyNhZ34N1JtRjwkn7ukjhuP3JuHjh84L2dBFCSH5Q0UWWnDqrHlUWLQ7ksMRwcMyDxsqSnJe4MUotFNd8BPGxbsQGj+Z0LCA5u2rdSgsOdI7jB8+dgkrJ4Qu/1wG5LPvU0EqLFuYSJU4PFEfR5fZFRO10AUBVmRYjk35RjzkfhyuIMqN61mtH6HTRvi5SSC5vGAbd+zcEZFyy60WdLkIIkRYVXWTJYRgGW9or0D/iziotLhiOYXTSj4YF5mFlQr5qG9jSaoQP/xp8LPe7ydvWVcIXjMLlC+OPP9IBkz63AoVhGHQ0mtF1wYlYPPfUx1zE4gn4glFRO11AsrC0O4OIxqT//oSi60pqpQwlGjklGJKCcvnfH4ws0KnltKeLEEIkRkUXWZI2r64AwyCrQI3z417wABorxSm6GJaD8vr7wXsnETn1Ss7Ha6svxY0dNjy0czWaRBgCDSQDSMKROPouukQ5XrY8/plN/jpxi64qixbxBC95wcPzPByu0JxFFwCUl1KCISksly886/2l1ygovZAQQiRGRRdZkkx6JVbXl+Jg5zgSGe6lEvb+iNXpAgBZVRtk9RsRObELCX9uM6NYhsEDd7Zi06pykc4umfAn45iC7+ty+4UZQmIvL0wGV0i9xNDjjyAcjc8K0RBYTWra00UKJhSJIRyJz93poqKLEEIkRUUXWbK2tFdgyhNC71Bm3ZvBMS/KTerUBnOxKDd/HOATCB9+RtTjikGlkKG5xojTOc43y5UrNRhZ3E5XRakGLMNIXnQ5XCEAQJlx7gAWq0kDty+CUCQm6XkQMhe3MBj5yk6XWg5fkII0CCFESlR0kSVrfXMZVAoO+zvHMvq6c6NuNIrY5RKwJWVQrPkQYv0HEZ/oF/34uVrTaMbopB+T7sItfxM6XWLv6ZLLWJSb1BhxSDury+5KdrHmW15oLU2GadhpiSEpgPdvalzR6dLIKUiDEEIkRkUXWbKUcg6bVpXjaI8D4Uh6A2md3jBcvggaRNrPdSXFujvBaIwIHfgFeL6woRVX6mgyA0BBu13CnfgSkYsuIJlgOCpxp8vuDIIBYDHMv7xQeBwh+eZKdbpmLy+MRBM0uJsQQiRERRdZ0m5or0A4EsexXkdajxf2c0nR6QIARq6C8rqPIeEYRKx3vyTPka2KUg0sBlVBo+Pd/gh0ajlknPgfTVUWLeyuIKIx6S4sHa4QTCVKyGVzn7+w14v2dZFCEDpds5YXapL/30/dLkIIkQwVXWRJW1ljhMWgwoE0lxieG/OAYxnUWnWSnZNsxWaw5U0IH3kWiXDxXHwzDIOOJjO6LkznJVp9Lu45ktXEUmnRgueBsSnpfuYOVxDl8ywtBJJ75wxaBSamqdNF8s/ti0DGsdAoZZf9uX5m/yolGBJCiHSo6CJLGjszs6vrvBPTntCijx8c9aCmXJfTsOHFMAwL1ZZPgA95Mfna45I9TzbWNJoRiSbQW6DoeJcvAqMESwuBZKcLAEYc0i0xtM8zo+tSlGBICsXlD8OoU8wa3K3TzBRdFKZBCCGSoaKLLHlb2ivAAzh4ZuGZXYkEj8Fxr2jzuRbClTdCsX4nfKd2IzpwWPLnS9eqOhNkHIvTBYqO9/jDszb5i8VaqgHHSpdgGI7E4fFHFi26aFYXKRSXNzxrPxeAVFIrxcYTQoh0qOgiS165SYOV1QYc6BwHv8DMrtEpP8KRuKjzuRai2HA3lFUtCO19AgnvZF6eczFKOYdVtUacKsC+Lp7n4fZHRE8uFMg4FhWlGsnCNByuZCE134wugdWkhscfQTBMsfEkv9z+yKz9XMD7e7oowZAQQqRDRRdZFra0V2BsKoDz4955HzM46gGAvHS6AIBhOZTf86cAzyO0+z/BJ4ojOayj0Yzx6QDsrvx2Y/yhGGJxXrJOF5Dc1zUsUWy8UHQtvryQYuNJYbh8c3eSNUoZGIY6XYQQIiUqusiycM0qK2Qci/2n5w/UODfmgVopS81Syge50QrVjZ9GfKIPkeO78va8C1kjRMfnudvlFmYISdTpApL7uibdobRHCGTCnm7RNfP6on1dJJ/C0TiC4ficnS6WZaBVyeGjThchhEiGii6yLGhUMmxotuBw1wRi8bmT+QZHPWi06cFesclcavKVWyBbcT0ix15AfLwvr889F2upBuVGdd73dQmDkee6KBRLVVkyTGN0SvwlhnZXEBqlLLU/Zj6p2PhpKrpI/rhTcfFzd5L1NCCZEEIkRUUXWTa2tNvgD8Vwsn92MRGOxjHs8Es2FHkxqhs/DUZnRnDPf4KPFP5ivKPJjO4LTklnWl1JysHIgsqZBEMp9nU5XEGULbKfC0jumzPplRSmQfJKGIw830gGvVoOX4DSCwkhRCpUdJFlY3WDCQatYs6ZXRfGvUjwPBpthgKcGcAo1FBvfxi8bxqhvU8uGPiRDx2NZkRiCfQM5S863uVf+E68GMpNasg4RpLYeIdz8bh4gdWkpj1dJK9ci3S6dBoFdboIIURCVHSRZYNjWWxebcWpgSl4r7ije24mRKNQnS4A4KwroNh4D2IDhxDrO1Cw8wCAVbVGyGVsXlMM3b4IFHIWKoV0M9I4loXNrBU9Nj6R4DHpDi04GPlS5SYN7ekieSV0uuYtutRyCtIghBAJUdFFlpUb2m2IJ3gc7pq47M8Hxzwwl6gkDXFIh2LdDnAVzQjt/zkSHnvhzkPOYVWtKa/7uoS4+CsHt4qtyqLF6KS4CYbT3hDiCX7RuHiBtVQNbyCKQIhi40l+uH1hyDgGWpVszr/Xa5JBGoXushNCyFJFRRdZVqrLdai16nCg8/JByedGPQXtcgkYloVq+8MAwyK4+z/AJwp3Ub6myYwJZzBvHRn3PHHWYqu0aDHlCYs6J8sxs1SwzKBK6/HlRkowJPnl8oVh0CrnvamhU8sRT/AIhotjdAUhhCw1VHSRZWdLuw3nx72pJWZufwRTnhAa8zQUeTGszgzVTf8bCfs5RN57oWDn0dFYCiB/0fFSDka+VJUEYRqpuPgMOl0AFV0kf1y+uQcjC4TUTV+QwjQIIUQKVHSRZWdzmxUsw+DAzMyufA9FToe88VrIW7YicnwXYqPdBTmHcpMG1lINTuVpiaHbF4FRK32nS4iNF3Nfl8MVAscyKNWn2+lKFl32aQrTIPnh9kcWDKnRa5IFmZf2dRFCiCSo6CLLTolWgY7GUhw8M45Egse5MQ9YhkGdVV/oU7uMcssnwBjKEdrzE/AhcfcgpaujsRQ9Qy6Eo9IuOYpE4wiEYyiRcEaXwGJUQyFjRe90WQwqsGx6+9EUcg6lJUrqdJG8cXnD88bFA8k9XQAowZAQQiRCRRdZlm7osMHli6DrwjQGR92oKtNCKWFqXjYYuQrq7Z8HH3AjtPeJgmxwX9NkRjSWQM+QU9LnSQ1GzsPyQpZhkgmGDvEKWYczvRldl7KaNDSri+SFcFNjoU5XankhdboIIUQSVHSRZWntCgs0Shn2nx7H4Ji3qJYWXoora4DimnsRGzyKaM87eX/+lhojFHIWpwemJX0eoejKR5AGkFxiKO7ywvRndAmsJjUmpqnTRaTn8i88GBm4dE8XFV2EECIFKrrIsiSXsbi2zYojZycQCMfQUCQhGnNRrP0guMpWhA/8AgnX7MHOUpLLOLTWmnDq3KSknTb3zODWfEX2V1m0cPki8Idyv8D0BaMIhGNpz+gSlJs08IdidJFLJOdeZDAyAKgUHGQcAy8FaRBCiCSo6CLL1g3tFRDqiGLtdAEAw7BQ3fIQGE6B4Jv/AT6e34v0jiYzHK6QpEvhUssL87CnC0jGxgPAiCP3bpdjJrkw06KLEgxJvrgXGYwMAAzD0IBkQgiREBVdZNlqrCyBtVQDpYJDpVlb6NNZEKs1QbntASSmLiD87nN5fe6ORjMA4JSE0fEuXwQM836CmtTEjI0Xiq7MlxcmZ3XZaV8XkZhT6CQvclNDr1FQeiEhhEiEii6ybDEMg0/cthIf374i7dS5QpLXb4C89RZET72C2PCZvD1vmVENm1mD0xJGx3v8YZRoFHn7dyg1qKBUcKLs6xKKpkyLrjKjGgwD2tdFJOf2RcCxTGrf1nx0ajktdyWEEIlQ0UWWtfZGM7atqyr0aaRNef3HwRorEXrrUSSCnrw9b0ejGT1DToQj0kTHu3z5GYwsYBkGlWatKJ0uuysIg1aRcfqlXMbCXKKiTtcyl+B5RGPSjmRw+ZJx8Syz8E0NvUZOkfGEECIRKroIuYowMiVUt34efMiH0Os/Ah/Lz6b3NU1mxOI8zkoUHe/2RfKWXCiosogTGz+ZRXKhwGpS056uZe71dy/ir358EPFEQrLncPvCMKQxeDy5p4uCNAghRApUdBFyleHMtVDd8iDi430Ivflj8Alp75IDwMpqI5RyDqcl2tfl9i88uFUKlRYtPIEovDleZNpzKLrKTRpMTAcLMoONFIeDnePw+CNwuEKSPYfLH0krpEanliMQiklaABJCyHJFRRchVyF503VQbvkEYheOI5yHwclyGYu2ehNOn5sS/bkSCR4efzSvywsBoLos9zCNaCwBpyeM8gwHIwusJjUCYYqNX64mXUEM2ZPd1rEp8ebGXcnlDS+YXCjQaxTgAfhDMcnOhRBClisqugi5Sinab4Ni/U5Ee/YikodEw45GMybdIYxNibsczheMIsHzaV0UikmIjR/OITZ+0h0ED6DMqMrq68tLkwmGUsbxk+J1vG8y9d9iv68E0VgC/lAsrU6XXpMM2qAEQ0IIER8VXYRcxRSbPgL5qpsRObELkdOvSfpcQnS82CmGrjwPRhaY9EqolVxOna73Z3Rpsvp660yHjBIMl6djvQ5UlWlh0Ckk63SlBo+ncVNDSDekfV2EECI+KroIuYoxDAPljZ+GrH4jwgefRrT/kGTPZTaoUGXRij6vyzMzGDnfe7oYhkGVRZdTbLywD6csy+WFqdh46nQtO95ABL3DLqxfWYZKs1ayTpcrg8HjqaKLlrsSQojoqOgi5CrHsCxU2x8GZ2tB6K1HERvulOy5OhrN6Bt2IRIVL7zD5ZspuvLc6QKSSwxHJ/1Z71OzO4NQyjmUaBaefzQfGcfCYlDBTgmGy87J/inwPLCh2YIKswZjUwFJ9ma6vMlOV7p7ugBQbDwhhEiAii5ClgBGpoD6jj8Fa6pE8LUfIm4/J8nztNQaEYvzODcq3owwtz/95U9iq7Jo4QtGU922TDlcQZQZVWAWmX+0EOtMgiFZXo73OWDSK1Fn1aPSrEUwHIM7y9fhQtypTnI6ywtlAAAf7ekihBDRUdFFyBLBKDRQf+hLYNQlCP7uX5BwjYn+HCurDWAA9F50iXZMty8CtZKDUp7ZcGExVM4kGGa7xNCRQ1y8wGrSwO6SpstBilM4GseZwWlsWFkGhmFQYU7uCZRiiaHLFwbLMKmQjIXIZRyUCo6WFxJCiASo6CJkCWE1Rmg+/JcAyyLw8v9Fwi/uMGONSo6ach16RCy6XP4IStIY3CqFakv2RRfP86IUXeWlagTDcUqMW0bODE4jEktgfbMFAFBpTr4OpQjTcPmSM/DYNLuxerU859l1hBBCZqOii5AlhjVYof7QX4AP+xF8+Z/Bh8W9kGuuNWJgxI1YXJwBqh5fGMYC7OcCgBKtAlqVDCNZxMa7fBFEYomsZ3QJrCYhNp72dS0Xx3sd0KpkaK4xAkiGXKgUnCSdLrcvvcHIAr1GTnu6CCFEAlR0EbIEcZZ6qD/wJ0i4xxF85d/Ax8KiHbulxohILIHzY15RjufyR/KeXChIJhhqs4qNfz8uPseiq1SIjad9XctBPJHAif5JrGmyQMYlfwUzDAObWSNdpyuDTrJOraA9XYQQIgEqughZomRVbVBtfxjxiX4E33gEfEKcxMGVM3fney6Ks3TR7Y9kdFEotqqyZGx8pnuqhKIr1+WF5hIVWIahTtcy0XvRDX8ohg0zSwsFFaXSxMa7Mux06dRy2tNFCCESoKKLkCVM3ngNlDd+CvGhkwi984QoYQ0lGgUqLVpR9nWFIjGEI/GCdbqAZGx8MBxLRdeny+4MgmGS88tyIeNYWIwqmtW1TBzvdUAuY9HeYL7sz21mDZzeMILhmGjPFYsn4AtG04qLF9DyQkIIkQYVXYQscYq27VBsuBux3r2IHHlWlGO21BjRP+xGPJHbvq5UnHWB9nQBydh4ABhx+DL6Ooc7iFK9KrVELBdWkwb2aXG7HJ5ABPEEJSIWE57ncbzPgdX1pVAqLk/rtM2EaYyL+Dpw+zIfPK5TyxGOxBGNiTeLjxBCCBVdhCwLio33QN56CyInX0bk1Ks5H6+5xohQJI6hicwKlStlc1Eotmxj4x3OYM4hGgKrSY0JZ1C02Hi3P4K/+fFBPPNGryjHI+IYmvBhyhPG+pWWWX9nS8XGi7evy+VPfzCyQIiWpzRNQggRFxVdhCwDDMNAecOnIGvYhPChXyLauy+n4wmpa7nO63L5Zi4KC7inq0SjQIlGnnHRZRchLl5gLdUgHI2LNhx378lRhKNx/M/+c9SxKCLH+xxgGGDtHEVXuUkNjmVE3dfl8iZfT5kUXTp18gYI7esihBBxUdFFyDLBsCxU2x8GV9WG0FuPIXLqlayPZdIrUW5So2cot6IrtbywgJ0uILmvK5PY+GA4Bm8gijJjbvu5BFaTkGCY+wV3IsHjrRMjMOmVcPsiONQ1kfMxiTiO9U5iZbURJZrZr3cZx6LMqBa16HLPdLoyeX+lOl1UdBFCiKio6CJkGWE4OdR3/NlMx+tXCB38JXg+u31ZLTVG9A27kMhhSZzbFwHHMtCq5VkfQwxVFh1Gp9JPMEzFxc/M2MpVeakwqyv3MI2TA5OY9oRx/20rUVehxxtHh0VbtkiyZ3cFMezwYcMcXS6B2LHxLl8EDIM5i7z56GbeixQbTwgh4qKii5BlhpEpoLr1jyBffRuip19F6M3/AB/P/AKrucYIfyiW1WBhgdsfRolWAZZhsj6GGCrLtAhH4pjyhNJ6vFgzugTmEiU4VpzY+D3Hkl2udSst2Lm1CRftvpyXgZLcneh1AADWNZfN+xibWQu7Myja4HGXb+b9xab//tLNdLpoeSEhhIiLii5CliGGZaHc8gkor/sYYueOIPjyP4OPZHbB31Kb+74uty9S0ORCgZBgmO6QZIcrWZyJtbyQY5NLy+w5dromnAF0Dk5j27pKcCyLmzdWQ6eW47V3L4pyniR7x/omUV2mW7BQt5k1iCf4VFGfK7cvktF+LgDQqeRgAHgD4uwvJIQQkkRFFyHLFMMwUKz9MFS3PIT4eB8CL/4DEv70Bx5bDGqYS5ToGcp+SLLbn/lFoRSqhATDNLt2dlcQWpUMGpV4yyKtJjUmpnO72H7r+Ag4lsFNaysBAEo5h23rKnGib1K0C3mSOU8ggr5h16yByFdKxcaLtK/L5QvDmOFNDXZmuS/t6SKEEHFR0UXIMidfuQXqD/05Et5JBH77HcSdI2l/bXONCb0XXVnvGXLPLH8qNK1KDoNOkXaCocMZEC0uXmAt1cDuCmT9s4xE49h3agwbmssuK2RvWV8FhmHw5nvDYp0qydDJvknwPLB+5fxLC4H3Y+NHRdrX5faFYcjipoZOLac9XYQQIjIqugghkFW3Q7PzK0AijsALf4/YeHrznVpqjfAEolkNdI0nEvAGojAWOLlQUG3Rpl90uUKixcULyk1qRKIJuHzZLes6ctYOfyiG7RuqLvvz0hIVNq0qw95TYwhFYmKcKsnQ8b5JmEuUqLXqFnycWimDUacQpdMVi2f//tJp5LSnixBCREZFFyEEAMBZ6qC5+2/BqEsQ/J9/RHTw6KJfI8zr6sliX5fHHwUPZHUnXgqVFh3GJv2LpjHGEwlMecQvuqwzSYjZxsbvOT6MSos29W9yqds21SAYjmH/6fGczpFkLhSJoXNwGutXloFJIzDGZtZiVISiy+OPgEdmM7oEerWchiMTQojIqOgihKSwJWXQ3P01sOY6hF7/d0TOvLHg460mNQxaRVZhGqkZQkWwvBBI7uuKxBKYXGTv07QnjHiCFy25UJCa1ZVFguHgmAeDY97UUsIrNVWWoMGmxxvvDecU8U8yd2ZwGrF4AhsWSC28lM2swfh0+uML5iPMwMum6NKp5fAFKUiDECItXzCKaEyctNarARVdhJDLsCo9NDv+GrK6dQjvfwrhI/897wUgwzBorjGiZyjzfV1uX3EMRhZUziQYLrbE0D5TlInd6SotUUHGMVnN6tqMRnHKAAAgAElEQVRzbARKOYct7RVz/j3DMLhtUw0mpgPoPDed66mSDBzrnYRWJcPKGkNaj7eZtQiG41kvMxW4vJkPRhYIywtpvhshRErffuJd/OrNvkKfRt5Q0UUImYWRKaG6/Y8hX3UzIid2IfTWY+ATc+8Haqk1wukNY9Kd3owrgXAnvmg6XWnGxjucwmBkcYsulmVQZlRnvLzQF4zi8NkJXN9eAbVSNu/jrllVDoNOgTeOUnx8vsTiCZwamMS6FRZwbHq/boUwjfEcwzRcOXS69GoFYnEeoUg8p3MghJD5BEIxTLpDONQ1gWhseXzWUNFFCJkTw3JQbv0MFJt+D7G+/Qi+8m/gI7O7MKl9XUOZLTF0+4TlhcWxp0utlKG0RLlobLzDFYSMY2DUi3/eVpMm41ld+0+PIRpL4Jb1VQs+Tsax2L6+Cp2D0xgTKR2PLKzvogv+UAzr01xaCLwfG5/rvi6XNwwGQIk287EG+pkByRQbTwiRijDGJBiO4dTA8liBQUUXIWReDMNAueFuqG56APGRLgR2fQ+JwOXFVaVFC51anvG+Lpc/Aq1KBrmseD6GKtNIMLS7grAY1GDTCEXIlLVUDbsrmPa+qwTPY8+xEaysNqCmfOFkPADYtr4KMo7FG0cpPj4fjvVOQiFjsbqhNO2vMeoUUCm4nBMM3f4w9FpF2h22S+nUyaKLYuMJIVIR9i9zLINDXcsj5Kl4rnYIIUVLvuomqO/4UyRcYwg8/21EB99L7fdgGQYrqw3ouZjZkGS3L1I0yYWCaosOY1MBJBLzFz0OZ1D0pYUCq0mDaCwBpyec1uO7BqdhdwVxy4aFu1yCEo0Cm9us2N85Bn+ILqilxPM8jvU5sLqhFEo5l/bXMQwzk2CY4/JCXyTrcQy6mU4XhWkQQqQidLqub6/Ayf4pBEJLf6QJFV2EkLTIatdCs/OrYJQahF7/IYK/+2ck3Mm7Uy21JjhcIUx70t/X5faHi2Y/l6DSokUsnkiFZVyJ53k43EHRQzQEQoKhPc0Ew93HRlCikWNjc3naz3HbpmpEogm8c3I0q3Mk6bkw4YXTG150IPJckgmGOS4v9IWz2s8FJCPjAVBsPCFEMnZnECUaOW5eV4VYPIH3eu2FPiXJUdFFCEkbV1YPzUe+BeX19yM+MQD/s3+L8JH/RostWSxkssQw2ekqrqKrqmwmwdDhm/PvfcEoguG4dEVX6cysrjT2dU26gzg5MImtayszWqJZa9WjpcaI3e8NI55YPlG9+XasdxIMA6xbacn4a21mDZzeMILh7O/8unPpdKmTX0cDkgkhUnG4gig3adBg06PcqMbhrolCn5LkqOgihGSEYTkoOj4A7X3fhazpWkRO7IL5ne9ik2YYvUPpLTHkeR5ufwTGIgnREAjJcfPt6xI6YGLP6BIY9UrIZWxas7rePpHsVG1bV5nx89y2qQZTnjCO905m/LUkPcf7HGipMab2R2VCCNPIttsVTyTg8UeyDqlRKzlwLEOdLkKIZOyu5KoRhmGwebUVZy844fKlt7T+akVFFyEkK6zGCPUtD0G98ytgFGp8SrUbay8+nVpyuJBgOIZoLIGSIlteqFLIYDGo5o2Nd6RmdKkkeX6WYVBuUmNieuFOVzSWwN6To1jbZIHFkHkBuH6lBRaDiuLjJTLhDGDE4c9qaSHwfvG/2PiC+Xj8UfBA1p0uhmFmZnXRni5CiPiisTicnnBqf/R1bVbwPHDk7NJeYkhFFyEkJzJbCzQf+RbOVX4QtsQ4fDNLDvno/Hes3KkZQsVVdAHJeV3zxcYLM7qkWl4IJLtoi3W63uu1wxOIYnuaARpXYlkG2zdUo3fYjQvj3qyOQeYndBDXZ7G0EEi+vjiWybrT5fYn33vZ7ukCkvu6qNNFCJGCwxUCj/dXjdjMWtRV6HHozNJOMaSiixCSM4bloFv/Ify9+x54LGsQObEL/me/elnK4aVcvuIajHypyjItxqcDiMVn73eyu4Iw6hRQZJBGlylrqQYOV3DBBMU9x0ZQblSjLYMo8ivdtNYGpZyjbpcEjvc5UFuugyXL4lzGsSg3qbPudLm8M++vHIounVpOe7oIIZIQluqXXZIEvLnNivPj3pxDhIoZFV2EEFHUVegRlmnxtvaDqSWHV6YcClKDkYssMh5IdrriCX7OMAuHKyRplwtIJhjG4vy8SZAX7T70Dbtx8/qqnGaFaVRy3NBRgcNnJ1KdR5I7tz+C/mF3RgOR52Iza7O++HClOl3Z39TQaRRFWXQFwzH0DDnnvJlDCLk6CKtGLh2/cm2rFQywpLtdVHQRQkQh41isqDKgZ8iVWnJ4ZcqhsOSwuJcXJocMz9VlcLiCkoVoCKymhRMM9xwfgVzG4sY1tpyf69aN1YjFebx9fCTnY5Gkk/2T4JH90kKBzayB3Rmcs+O6GJc3DAbIac9kMS4vfK/Hga89egjff/o4Hnm+syiLQkLI4uyuIFQKLjWeAgBMeiVW1ZlwuGtiyd5UoaKLECKalhojRhw++ILROVMO/c98BZGu3fB6A5BxLNRKWaFPeRabWQOGmR0bH4nG4fSGL1sOIYX3Y+NndzmC4RgOdo7j2tbyrFLxrmQza9HRaMae4yOIxig+XgzHeh2wGFSoKdfldBybWYN4gk+Ft2TC7Y9Ap5FDxmX/K16nlsMfjC64zDVfpj0h/PC5U/j3509Dr1HgzuvrcKJ/Et98/EhGYyoIIcVBuIHJXLFa47o2KyacQZxfonuN0/pEHhwcxH333Yc77rgD9913H86fPz/rMc899xx27tyJu+++Gzt37sSTTz4p9rkSQopcc40RPIC+4fcvhC5LOdQaEd73JLYO/gh36LuBWPEta1PIOZQZ1bNi4yfdyeV+Ui8vTO4ZY+dMMDzQOY5wNI7tG6pFe77bN1XD7Y/g3e6lPyNFasFwDF3nndjQXDbrYiJTQmz86GTmSwyTM7pyW7qr18jBA/CHCtdNSiR4vPneMP72scM4MziNj97chK9/ZhPu3daEr35qI2Qsi+8/fQwv7BukmXOEXEUmnME5b2BuaimDjGNw6MzS/H2UVtH1jW98A/fffz9effVV3H///fi7v/u7WY+544478OKLL+KFF17AL3/5Szz++OPo7u4W/YQJIcWrsbIEMo6d8+6zzNYCzd1fh/rOv4aTMeID7CH4f/mXCB/fBT5SXBtnqyzaWcsLpZ7RJWAYBuVGzaxOF8/z2HN8BPUVejTYSkR7vtUNpbCZNXj96PCSXdKRL2cGpxGLJ3JeWggAFTMdz/HpzMM0nL5wzoPHdZpkJ7VQS/iG7T5896n38IvXe9FUWYJvf+46fGhzXap712ArwTc+ew02t1XghX2D+Kenj2PKPfc+SEJI8UgkeEzOs1Rfo5JjTZMFR85OFEWXXWyLFl1TU1Po6urCjh07AAA7duxAV1cXpqenL3ucTqdL3dkLhUKIRqM53+kjhFxd5DIOjZUl6Bmae8kPwzCQVbXhF/xOvGz4ONiyBkTe/W/4nv4Swkd/Az7km/Pr8q2qTIuJ6eBlS+5ScfESLy8EAGupGvYr9nT1XnRhdNKPW7KMiZ8PwzC4bWM1Lox70T/iFvXYmeocnMKf/mDvVRvscazPAZ1ajhXVhpyPpVbKYNIrs+x0hXPvdKmTRVu+93VFonE89/YAvvXEu5hwBvHgzjb8xX3r5rxAUytleHBnGx7c0YYLdh+++fgRvNeztOf8EHK1m/aGEE/wl4VoXGpzmxVufwRnh5x5PjPpLbqhYmxsDFarFRyXjEjmOA7l5eUYGxtDaenlccVvvvkm/uVf/gVDQ0P40pe+hJaWloxOxmzObQ28FMrK9IU+BbLELbXX2PqWcjz7Zi+0ehU0qrn3HXkCEbDNzai796MIj52D68Bz8B97EdHTr6FkwwdguO4uyPSmPJ/5+1Y1WLDrwAWEeaBy5t/HF4lDreTQWFsq+Q2lhiojTvZPorRUC27mzv5//a4bOrUcH97aBJUi/b1w6by+7rp5JX6zdxDvnB7HlvU1WZ93rs68NQBvIIrzdh9uu7auYOeRjVg8gdMDU7i+oxIV1tyLLgCoqyjBpCeU0WdEPMHD44+gslyf02dLTTgOAGDlsgWPI+bn18leB/79uZMYm/Tj1mtq8MDO9rTCQO66RY9NHTb801Pv4d+f78QHr6/HH9y1OqP3CSlOS+33o1RefGcAx3rs+PJnrin61/2oK9mRXllvnvPf91ajBk+80o2TA9O4+Rppfw/k+/Ul6r/MrbfeiltvvRWjo6P4whe+gJtuugmNjY1pf/3UlK+o2ollZXo4HEtzMx8pDkvxNVZt0SDBA4dOjqCj0Tzr76OxBLyBKJQck/zeZWVgb/o8NB07ETm+C+4jL8F99GXIW26CYu2HwepzX6qVKb0yeZOps88OnTxZ9FwYdcNiUGNyUvpunF7JIRbn0X1uEuVGNVy+MA6eHsOtG6vhdQeR7ismk9fX1g4bXnv3InoGHCgtUWV/8jk4PtOlOHByFGtzmEFWCGfOT8MfiqG11iDae9qsV6L7wjTsdk/ahb7bF0aCB+QMcjqPWDjZ4Rqd8MBRMfcNUbE+v7yBCJ7Z3Y/9neMoN6nxVx9fh9b6UoQDYTgC8w9Zv5QcwF9/fB2ef+ccfnfwPE73OfDw3atRXVZ8N3NJepbi70cp9I+48diLneB54P/98hge+HBroU9pQX3npwAACoaf9993w8oy7D81go9ua4BcJs1cTCleXyzLLNhAWnR5oc1mw8TEBOLx5F2veDwOu90Om23+uOLKykp0dHTgrbfeyvyMCSFXtRWVBnAsM2+qmCcVF3/58ifOVAX19oehve/7kK+8AdHut+H/1d8g+NZPkXDld25HRakGLMNctq8rH3HxAiHB0D4zp+mdk6OIJ3jcsl7cpYWX2r6xCjx4vHlsWLLnWMi0JwS7MwiFjEXX+em8BiPwPI9/fPoY/u6nR/Dkqz040DkGuzOQ0R63Y70OKOQsVteLVyzaLBqEIvHUMPF0CI/NdXmhkI7pDUi31JPneRzsHMfXHj2MQ10T2LGlDt9+4Fq0ZvkzlHEsPnrLCvzFfWvhDUbxnZ8dxe5jtFeRLF3BcAyPvnQG5hIVbt9Ug32nxrD/9FihT2tBdmcQHMugVD//zb3rVlsRDMdxsn8qj2cmvUWLLrPZjNbWVuzatQsAsGvXLrS2ts5aWjgwMJD67+npaRw+fBjNzc0iny4hpNgpFRzqK/TomafoEga3zrdsiC0ph+qmz0L78X+EfPV2xAYOwf/sVxB888eI2wfycgEll7Gwlqox4kgWXQmez8tgZIGw1n3CGUQ8kcDbJ0axuqE0VYxJwWJQY0NzGd45MYpwNC7Z88yne2b9/m2bauAPxTA4lr873BftPnQPucCDx+GucTy26yy+/J+H8Oc/2o8f/eY0Xjk8hP5hN6KxuX8uCZ7Hib5JdDSYoZCLd1fWNvPvPTqVfpiGy5f7YGQgmeKplHOS7emyOwP4l1+fwKO7umA1qfGNz16Dj9zUJMrPr73BjG8/cC1aao146rVe/Og3p6/qmV7TnhC6zk8v/kCy7PzyjT5MukN4cGcb7tu+Aqtqjfj5az2z0neLid0VhMWoBsvO371vrTXBoFXgUNfSSjFMa3nhN7/5TXz5y1/GI488gpKSEnz/+98HADz44IP4kz/5E3R0dODXv/419u/fD5lMBp7n8clPfhI33nijpCdPCClOzTVGvPbuRYSjcSivuIjy+NIbjMzqzFBt+QQU63YgevpVRLp2IzZwGKy5BvLWWyBfcT0YhXRFUKVFi4v25FJClzeMWDyRlxANADBoFVAqOExMB3CibwpObxifvF36m1i3b6rBez0OHDwzjpvXSddVm0v3BRe0KhnuuLYGvzt8AZ3nprCiSpy9UYs5OZC8m/qX962DXqPA6KQf/SPu1P+O9ToAADKOQV2FHiuqDFhRZcSKqhIYdEpcGPfC6Q1j/TZxl8LaLMnY+PGpQNodNPc8neRs6NRySYqVY70O/OTFM2BZBp/8QDNuXl8FVuR9kiVaBf7so2vxxrsX8exbA/jGfx3BQzvb0FJbuL2i2QhH4/jnX5/A+FQA//RHWwq29JcUn6Pdduw7PYYdW+qxstoIAHjortX45uPv4se/7cTXP70JSoU0S/Ny4XAGYV3kdynLMri21Yo9x4cRCEXn3R9+tUmr6GpqasKzzz47688fffTR1H9/9atfFe+sCCFXtZZaI353eAjnRj1orbv8Isc1c1Fo0KZ3UchqDFBe9zEo1u9EtP8QomffQnjfkwgf+jXkK66DvPUWcGUNon8PVRYtjvU4EInGUwNq87W8kGEYWI1qTDiDGJ3yo7REiTUrZu+PE9vKagNqrTq8cXQY29ZW5jWBtnvIiZZaE/QaBRptJegcnMY9W9PfE5yLU/2TaLDpYZgpVKrLdagu1+HmmeWcbn8EA5cUYW++N4JXj1wEAJQZVVDKObAMgzVN4hZdBq0CaiWXWafLm+x05RoZDyRj46UouvYcH0GJVoGvfHIjTPrci8P5sAyDD1xbi5ZaE/7jhU7849PHsWNLPe7e2iB6kSeVZ/f0Y2wqucx43+kx3HWD+J91S820J4STA1O4eV1+P8PyyekN42evdKPBpsddN9Sn/tyoU+KhnW3451+dwM9f68Ef3NlaVD8DnudhdwWxssa46GM3r7bi9aMXcbTHgZvWVubh7KRX3BEnhJCr0ooqIxgG6Blyziq63L4wGCSHr2aCUaihaLsF8tabkXAMInr2rWQR1v0OWEtdsvvVdJ1o3a+qMh14AGNTgdSMrjJj/u4yl5dqcGZwGsFwDL93UyM4Nq2xijlhGAa3b6rBT//nLLrOO7E6T2EWk64gJt0hfOCaZHJie6MZL+4bhC8YTe0tkorHH8G5UQ/uvnH+i1mDVoENzWXY0FwGIBkGc2HCi/5hd6oY27SqTPRzZRgGNrMW41Ppx8a7/BHo1PLUPKtc6NVySZYXjjh8aKsvlbTgulRdhR7f+Ow1+MVrvXjpwHlUlGpwfXtFXp47Fyf7J7H72Ag+cE0NLtp92Hcq2dW4WgrGQnnqtV6c6J9EW70JVpN0S7ILJcHzeGxXF6LxBB7cuXrWe72tvhR33diAF/YNoqXGiK1FVLB4g1GEIvG0bmDWV+hhNalxuGtiyRRd0v8WJ4QsOxqVDLXl+jnDNNz+CHSa7C8KGYYBV94I1bYHoPvkv0F5w6eARALhvU/A94s/R2jvE4hPXsj1W0DlzNKukUkfHK4gWIbJ69Ieq0mNYDgGjmVw05r5g4vEdm2rFSUaOd46PpK35xTmsayaKdDbG0vBIzlsWGqnBqbAA1i7Iv0ulVzGYkWVAR+8rhZf+EgH/vWLN+Lzd7dLcn62Uk3Gna5c93MJkp0ucYM0/KEoXL4Iqsq0oh53MSqFDA/c2QqLQYUDZ/IbzJMNtz+Cx18+i+oyHe7d1oita2yYdIfQc2HpzS4S07lRD070TwIABgo8d1Aqr797EWcvOPH7t65MDVG/0s4t9WirN+Gp13tTy+SLgT2DeZcMw+C6Niu6Lzjh9KaXYlrsqOgihEiiucaIgVHPZQOGAcDti6S9tHAxjEIDxepbobn329Dc/beQNWxCtPcAAr/5BvzPfxvR7nfAR7P7sLaa1OBYBiOTfjhcIZgNSlG6B+k/f/KX6caWstSyt3yQy1hsaClH5+D0rH87qXRfcEGvkaNqptBtqCiBViVD5znpk6tODkzCqFOg1lqc0eI2ixZuXwSBUCytx7v9YdFeL1Ls6RLCaaos+f95MwyDzaut6Do/DbeveC/ieJ7H4y+fRTASx8N3tUEu47ChuQxqpQx7TxV3Ml2h/XbvOejUcigVHAZGPIU+HdFdtPvw3NsDWL/SsmD3h2UZPLhzNTRKGR75bSeC4fQ+P6TmcGa2VH/z6grwAI6cXRqBGlR0EUIk0VJrRDSWwPnxy3/xuf3i3YkXMAwDzroC6ps/B90n/xXKLZ8AYiGE3vkv+J76M4T2/RzxiX7wfPpFhIxjUWHWYNThh90ZzFtyoaCxsgRKOYfbr8n/sOI1TWaEo/F5Y//FxPN8aj+XsPeAZRmsbijF6cFpJCRMq4zFE+gcnMbaFZai2vdwKSHBcHw6vSWGLl9EtPeXXqNAMBwXtfgWUtWq89zpEmxuqwDPA0fO2gvy/OnYfWwEpwam8LFbVqBqZs6YQs5h82orjvY44A9dvUmMUuq96ELn4DQ+vLkOjbaSJdfpisbi+MlLZ6BVyfG/P7Rq0c8sg1aBz9+9GnZnAE++2lMUoxPsriAYpL9Uv6JUg/oKPQ6doaKLEELmtbI6mTzXM3T5hbvLF4Fhnrh4MTBKLRTtt0Pzv/4e6ru+ClndOkR73kbghf8D/1N/huBbjyE6eBR8JLjosaos2plOV/5mdAkqLVo88hc3oakyPwl+l2qtM0EuY3FyYFLy57K7gnB6w2itvXxjdUejGR5/BMMSLo3puehCOBLHWpEDMMQkJBiOpbHEMMHz8PgjoiQXAsk9XQBE7XaNOHxQK7m87ee6UqVFi1qrDoe6inOJ4cikH8/s6ceaJjO2b7g8QXTrGhti8QQOL7EYbbH8du85GLQK3LKhCiuqDLjo8CEUKY4OjxiefWsAIw4/HrizFXpNer9DW2pNuGdrIw53TeDtE6MSn+Hi7M4gTCXKjAYeb15dgQsT3rQ+A4sdFV2EEEnoNQpUlWkv65YIF4X5WC7HMAxkFc1Qb38Yuk/+P6i2fx5cZRti548h9PqP4Hvyiwi8/H8R6XwdCa9jzmNUWrSYdIfgC0bzFhd/qUJ1X5RyDqtqTTg1IP3yvu4Ll+/nEgghHqclXGJ4sn8SchmL1vrijREvM6rAsUwqwW4hvkAU8QQv2k0NnSRFlx9VFl1BO4ub2yowOObFRJrdw3yJxhL4yYtnoFJw+OyHZ6fO1Vn1qCnX0RLDOZw9P43uIRc+fH0dlHIOTVUG8DzyOu9PSp2DU3jj6DBu3ViNjsbMkmzvvL4O7Q2lePqNPlwYL+zPI5sbmNe2loNhsCRuNlDRRQiRTHONEX0jbsQTyeVJ/qC4F4XpYpRayFdshvrWz0P36R9CvePLkLffBt47ifCBX8D/y7+C/9mvIXzkWcTH+8DPnK+wxwgAygz5L7oKaU2TGXZnMO1lbdnqHnLBoFXM2hBu1ClRU65D5zlpwjR4nsfJ/km01plmzZIrJhzLwlqqSesu7/uDkUXqdM0kjPoC4oRp8DyPkUl/3kM0rnRdmxUMgINFFqjxm3cGcNHuw2c/3DrnZyTDMNi6xoYL414MTSyNYkIMPM/j+b2DMOmVuHldcp9TU1UJgKURpuENRPDTXWdRZdHiozc3Zfz1LMPgczvboNfI8eMXCru/y+4MZLxU36hTorXOhENnJopiiWQuqOgihEimpcaIcCSOoYnkEjFhcKsYM4SyxbAcZJWroNr8cWjv+x60930Pys2/D0ZdgsjJVxB48e/hf+pPEdzzKGrDvVAiec7lBeh0FdLapuTd1FP90i0x5Hke3RecWFVnmrPz0d5Yiv4RtyQXCWNTAThcoYxSCwvFVqpJq9Pl8ok3GBl4v9PlFanT5fFH4AtGU8mghWLSK7GqzoRDXcVzEXfm/DRePXIRt2yowroFXpObV1dAxjHU7bpE5+A0+kfc2LGlPrVsTauSw2bWXPVFF8/zeOJ33fCHonhwZxsUWd4gKtEo8PBdqzHpCuHx33UX5HUfDMfgCUSz+l16XZsVdlfwqu9cUtFFCJFM88wARGFfl9snDEYuXNF1JdZQAcWaO6DZ8TfQffoHUN36h+Cq2xEbOgHVocfwD6Zn8Mf6V1E6+Cpiw2fAR0OFPuW8sBjVqLRocVLCJYbj0wG4/ZFZs9wEHQ1mxBN8agmimIT9akJxWcxsFg3sziBi8YUDLdypTpdYkfHJ44i1vHBYCNEocNEFAJvbrLA7i+MizheM4rFdXbCZNfjYLSsWfKxOLceG5jIcOjOOaCyepzMsXjzP4/l3zsFiUGHrFaM1mioNGBj1FE1hnY29p8ZwvG8SH7mpCbVWfU7Haq4x4iPbGnG0247dx/I3EkTgmJl3WZ7F7LSNzeWQcSwOFVl3OlNUdBFCJGPUKWE1qVP7usRe/iQ2RqmFvOm65D6wT/0A6p1fwTF2DTRcHPzp/0Hw5X+C74k/gv/5byF06FeInj8GPlQ8M1DEtqbJjN6LLsmWo5wV9nNdEaIhWFFtgFLB4bQE87pO9k+hplyX19lr2bKVapHg+dSMm/kI7y+xOsk6tQwARBuQnIqLLyt8PP/GluK4iBM6Gb5AFA/ftTqtpa43rrHBH4rheJ/0QTfF7kT/JM6Pe7FzS/2skR5NVSXwBaOYWOR9U6wmpgP45Rt9aK0z4QPXipNi+8HrarGmyYxf7+6blSwstVTRlUUolUYlw9oVZhzptqe2K1yNqOgihEiqpdaI3osuJBLJEA2gsMsL08WwHGS2FoTa78G7jQ9C95lHoP7Ql6BYdycYTo7omTcQeu0H8D35x/A/+zWE9v4M0f5DSPikH+ibL2ubkp2mrvPSfE/dF5woLVHOu8ZfxrFoqzOh89yUqHerfcEo+ofdWLui+LtcQLLTBSyeYOjyR6BVyTJKBlsIx7LQqmTwiVZ0+aDXyFFSBJ3u1EXc2YmCXsTtPTWGY70O3Lst/U5GW30pzCXKZb/EMMHz+O3eQZSb1NjSUTHr75uqksmvV+MSw1g8gZ+81AUZx+AP7mwFK1LwDMsw+NyONpRoFXjk+U4E8jh+wD5TdGU7fmVzmxUefyR1s+5qJCv0CRBClrbmGiPeOTmGYYcPLl8ESjkHleLq+ei58/r61H/Lajogq+kAAPCxCOKOQcTHexEf60G0/yCiZ/cAABh9GbiKZnC2ZsgqmsEYKop2DtRCmqoMUCtlODkwhY0t5aIeO8Hz6B5yYWHFpRcAACAASURBVE2TecGfTXujGcf7JjE+HYDNLM6ytM5zU0jw/FWxnwtAKmRksX1dLm9Y9C6yTi2HNyhOkMbIpP+ycJpC29xWgfd6HDh73on2DBPhxDA+HcDTb/Rm3MlgGQY3dNjw0v7zmHQHYVlmIT+C93ocuGj34cGdbeDY2T2ESosWaiWHgRE3buiwzXGE4rXrwHkMjnnwh/e0i96N16nl+Pzd7fj+L47hv17uxhd+rz0vv5/sziB0ajk0qux+/69pMkOtlOHwmQm0N1wdN8yudPVc+RBCrkotNcn9Oj0XXXD7w1dFlysdjEwBma0FMlsLsH4n+EQciemLiI/1ID7Wi/jFU4j17UcYAKPUgbU2gbOuSP6vrAGMvPiXtck4Fu0NpTg9kCxSxLrbCgCjDj98wShW1S4c194+Ex3feW5atKLr5MAU9Bo5GmwlohxPaiqFDKUlykU7XW5/RPT3l04jF2VPV2ImufDGIrr4XdNkhkYpw8EzE3kvumLxZDy8nGPxuR1tGb+3bpwpug6cHsddNzZIdJbFK5Hg8du951Bp0eK6Vuucj2EZBo2VBvSP5HcZXa76h9146cB53NBegWtWiXuzS7CiyoB7tzXhmT39eOPoMG6/RpzliwuxO4NZd7kAQC7jsLGlDEe77fhUNJ51qEghUdFFCJGU2aCCuUSF3osu+ALRogrREBPDcuAs9eAs9UDHHeB5Hgn3GOLjfUhM9CM+0Y/I0EnhwWBLa8BdUogx+rKi7IataTLj3W47hia8qK8Qr0g5OyTM55p7P5egzKhGRakGpwenRLkwiCcSOD0whfXNFlGLSKmlk2Do9oVRsUgRmym9WoFpT+7hMdPuEMKReMHj4i8ll7HYtKoMh8/aEY7G8zo64IV9gzg/7sUf3dOe1aBoi1GN1noT9p0ew44b6q+q17IYDp+dwNhUAH94TztYdv7vvamyBC8dOI9gOAa1svgveYPhGH7y0hmYS1S4//ZmSZ/rjmtr0HvRhWf29KOpyoDGSmlvQjlcQayYWfKZrevbrNh3agwnB6YkK0ilRHu6CCGSE/Z1ufI0GLkYMAwDzlgJxaptUG37A2g/9l3oPv0jqD/4F8l9YSoton0HENrzE/h/9dfJmPrXfoDwiZcRG+8FHxNnSVeuOprMYACc6hc3xbD7ghMWgyqtpVHtjaXoGXIhEs09ra1/2I1AOIa1TVfH0kKBzazF2HRg3r1tPM/D5ZOg06WWixIZ/35yYeFDNC61ua0C4UgcJ/IYStEz5MTLBy9g6xobNuVw4XjjGhsm3SFJ97gUY/JfPJHAi/sGUV2mw8aWsgUfuyI1JPnq6HY9/XovpjwhPLRzteRFIsMw+IMdrTDqlPjxbzsRFuHzdT6xeAJTnlBOnS4AaKk1waBTFDwAJ1tUdBFCJNdcY4Q3EMXEdGDJdrrSwah0kNWugfKae6G586+h+8wj0Nz7bShv/DS46g7Ep0cQOfIMgi/+A3yP/yH8z38b4XefQ2y8D3yiMPHQJRoFGipLRI2OT/A8ei+6sGqeqPgrdTSaEY0lUimYuTjZPwWOZbB6Ztni1cJm1iAcicPpDc/5976ZweNi7+nSa+TwBqI5X3yPOJIpn4We0XWl5lojTHpl3i7iAqFkPHyZSY3fv21lTsfa2FwGjVKGfRIFaoxN+fHnP9yHExLO6svGgc5xTDiD+L2tDYt2+ITuzdUQpvFutx37O8ex4/p6rKjOrSOULq1KjvtvX4kpTwjnJPwZTblD4Pnc512yLIPrWq04NTAFfx5DQMRS/L1WQshVr+WSSHCxZggtBQzLgjPXgjPXAm3bAQCJoAeJiQHE7f2Ij/UicuJ/gOMvAQoNZNWrIatZA666HaxW3GVkC1nTZMYLewfh8UdESZ67OOGDPxRDa5pL4VpqjJDLWJw+N53z3puTA5NYVWu8KpYaXUrYzzY2FZhzY73Yg5EFOo0csXgC4Wg8pwCckUk/SkuUWW+ilwrLMLiuzYrX370IbyACvUa6zyee5/Hkqz1weiP46qc25hwoJJdx2LzaindOjsEfikKrkot0psnOxGO7uuAJRHGib3LBgc35FIsn8NL+86iv0GPdysXPSaOSo9KixcBo8Xe6Xtp/HjXlOuy8oT6vz7uyOvn7+dyYB6310tyMyjW58FKbV1vx2rsXcbTbjm3rqnI+Xj5Rp4sQIrlyozq17MmgXR7LC7PFqksgq18P5bUfhebur0H36R9CddsXIG/YiPh4H0Jv/xT+X/w5/M99HeHDzyA22g0+Ic0cLcHaJgt4AKfPidPt6k7t50qv6FLIOTTXGNE5mNvzTzgDGJsKYE2RXEBmwmZeODZeGIwsdidZp05eyOcaGz/i8KOqyJYWCja3WRFP8DjabZf0eQ6dmcCRs3bcvbVBtP0zW9dUIhZP4NCZCVGOJ3j54AUMjnlRolWgv4i6RPtOjWHSHcI9WxvT3gPbVFmCgRE3EkW4VFIQiycwNuXHmibzrHljUtOp5Sg3qnFewkHhwozBXDtdAFBn1cNaqsHhLnFf8/lARRchRHIMw6ClJnk3bamkF+YLo9RC3nhNcl/YJ/4Vmnu/A8W1HwWj0CJy6lUEd30Pvp/9MYKv/RCRs28h4RN37xUA1Fp1MOgUoi0x7L7ghNWkzihAoKOhFGNTAUy6sx90enJmX9rVEhV/qRKtAhqlbN4wDacweDyLUIaF6NXJ92su+7riiQTGpgJFFaJxqZpyHaosWhyU8CLO4Qri56/1YGW1AXdurhPtuHUVetRaddh7alS0Yw6OefDi/vPY3GbFrRuqMDrpFyXBMlfRWBwvHTiPpqoSdDSm35FpqjLAH4phYnrhIJpCGpsKIJ7gUV2gweH1Nj0GJRyWbHcGoZCzotwUYhgG17dZ0TPkEiXkJ5+o6CKE5EXLzFKybJK6SBLDMODMNVCuuxOanV+G7jM/gur2L0LetBlxxyDCe5+A/+kvJYc1H/oVPMdeQ3TgCGLDZxB3nEfCYwcf9oPnMxsGyzAM1jSacWZwCrF4boNk44kEeofT388lEJYVdg5mP6j5ZP8kKi1alIuwxCXfGIaBzaxZoNM1s7xQ7E6XZqbTlcNFt90ZRCyeKKoZXZdiGAabV1vRP+zGpCv7on4+CZ7HT3d1gWGAB3e2LZi2l42tayoxNPH/27vz4LbO+1743+ccLMROEiRBkBQlbpKohbJjRV5ix4ntRE7jrcm9deqbNL1Jk7lvpv0jve3b9G0nW9O+dfu+naZTz2SmSe/t1G3SJn7jJE5iO/EWy45s2ZJF7RIhiiTABQsJkgBBLOec948DQKS4ASRAgPT3M8MBCBwcHogQeH54nuf7i2FofOMjFam0gm8/cx4uuwn/7cO70Z2bejZa+dGul98ZxdRsEh8rYpQLQD4xr5pG7G7kz655bKvQBxMdXicmZ5KYjpcnwCkUTaCp1lKyhN5b93ugAXjzQnlHp0utuiZXE9G2dedBL5xWY9WeeG1FwmSBseMWGDtu0SPqp0ah+PuRGTmD9NlfINy/UviGAEwWCLNtwZcVwqRfwmyHZK+HVL8DUq0HQjKgr6sBr/aPYcA/XXTBtNDwRAyJpLJmf64bed1WuJ1mnL06iQ+sYx5/IpnB5ZEoPrwJ/WjKxeu2rTjFczqWgtVsKHnvGkcJphcGQtnkwgp9il+IW/d58NQrV3H8/AQeuGNXSff98qkALvun8Znf6C1LI+Nb93nwHy8O4Fj/GHY2Oza0rx+84sNYZA7/89GbYKsxotPrhCQErvin0VfBxM9kWsFPfz2Eve21Ra87anZbYTUb4AvM4K6+ljId4cb4gzEYZAFPthH6Zsv1LBwcmynL+r1gNAFPCaYW5njqrOhqceLS8BTuv7W9ZPstNxZdRLQpjAYJt+zZen01tgohBOT6Vsj1rTD1fQSakka9VUN4LDu6lYwDyTi05By0VDx725x+mZqDFovol8k4sDApUTJAqmvBHlcr7rWkMNqfxu7GWyEsrnV9ankxG2+9t331/lzLPb8DnW68cX4CGUUtet3DucFJKKq2JacW5njdVhw7M4a5+TSsN4QmRGPlaTyeG+manVv/J+CBcBwC+slvtWpwWdDT5sLx8xP46O07S/aJ/OTMPL7/sg/7d9XhfQebS7LPG9ktRrxndwOOnx/Hb93TBaNhfYX3hWuT+OVbftz7nrZ8uqfZJKPdY8eAv7KjRC+dDGAmnsIXHjlQ9GMlIdDZ6oSvCkbrVuIPxdHitm36eq6cnR4HhAAGR0tfdKmahlA0UdSU0EJ84TcPIr3BmRebjUUXEdE2JGQjDE4H5GRxJ+KapgGZJNSZENTJESiREahTfmjjF/CQJQqMvY34k9+FqHFAqm/Lf8n1OyDVt0IYVp8+emF4Ci0NtnX1azvQ4cYr74zCF5jOT1ct1OmBMGw1BnS1lrcBaDktTDDsuqHJaDSeLHlyIQBYzQZIQmxoTVcgFENjnWVTmw+vx237m/Gvz13CSDCGds/GRoyA62mFmqbhd+7fW9bm53cdasGbF4I4eTmMW/d5in783HwG3/nZBXjqrfgvH+xadF93mwu/emd0XR92lEIimcHPjg9hf0c9du8o7sOanO4WF350bBBz85mqS9AE9OmFxY7+l5LZJKO1wVaWdV3R2STSGbXk07q34lKF6nvlERFRxQghAGMNZPcOyO4dMC5oJfTiry/h+Gtv43+83wXr/ASUiB/pi68A+UbOAsLVBLm2BcLRAMnuhrC79UtHAxSjDVdGptf9iX/vzjrIksDZwcmiii5V1XDaF8HBLjdkaesuZb6eYLhM0TWbwu4dpe/tI4SA3Wrc0JquQDi+JaYVv3dvE/79F5dx/NxESYquN85PoN8XwSfu7SlJVPZqenfWwe2swav9o+squv79l5cRzUbZ31gc97TV4pdv+TE8EStZ6mIxXnjbj1gijd+8q3Pd++hqdUGDPn2u2nr0xRJpTM0msaOpstNvd3mdeOdKGJqmlfQDglA0l1xYvSPdm4VFFxERFWT/nh148pUATsm7ce/72wAAmqZCmwlBmfRDnRyBGhmBOj0OdfQCkF6cLKVJRvxPqwX2qAfzv/JC2BcUZg43hK0OQlr5z5K1xoCuVhfOXI3g43d3rbjdja6OzSCWSONQBdeklEJDbQ0MslgSpqFpGqbjyXWNHhbCYTGue01XOqNgYjKxJaYW2y1GHOx0440LE/gvH+jaUODF7FwK//7LK+hsceK+W9pKeJTLk4TAnX1e/PjYIMLRBBqKKPLevhTE62fH8eAdu5YtqvJBFP7ophddc/NpPPvGMA51uTf0sztbnBDQmyRXW9EVqHCIRk6n15mP5C/lhwS5uPjGEq7p2qpYdBERUUE89VZ46iw47Qvj3uyJpBAShMsDyeUBOm7Jb6tpGpCagzobhhqLQItFMHDpKsJjo7hZTiMz9A60xA1TWYSAsNZB2Oog2eogbPWQ7PqlsNVDstWhr8OFH/xqCNPxVMHxw6cHwpCEwIESrynYbLIkwVNnXRIbH5/PIKNoZZleCOjFyHqnF45F5qBqWsVPKAt1234P3hkI49JIFL0bCIz57gtXkEhm8Lsf2VvytMKVvO9gM358bBDHzozhkQJHhaZjSfzLs5ew0+NYsSlvncMMt7MGVwLT+HAJj7cQz58YwVwyU/DzWYnFbEBLo60qEwxHgtmiq8IjXQvDNEpadEUTkCUBt3PrTQcsNRZdRERUsL6uBrx0KoBkSoHZtPIaHSEEYLZBNtsgN+h9iX565hRmLWm8/+NHAABaJgUtPgl1NgI1FoYWi+gFWjwKdSoAdeQMkEku2u+dEOirrcH80y/A1OjRR8eyBZmw1ekBHxaHnsgo9KmEpwci6GlzwXZD+MRW1Oy2wp89ScuJ5np0lakHnt1qXLE/2FoCYX1UbitMLwT0Hm5mk4zj58bXXXSdHgjj+LkJPPS+XZua2NjgsmDfrjq8dmYMD72vY81iT9M0/MuzlzCfUvB7D+5bdb1WT5sLF4amSj71bDUz8RSePzGCW/Y0bjiVEQC6Wlw4cTEIVdMgbdJzKIQ/FIfdYix5Y/NitTbqQR6DYzM40lv8FNWVhKIJuJ01W3pqd6mw6CIiooL1dbvxi7dGcGFoCjf1FD5dL51RMRCYxt03XY9sFgYThKsZkmv5NV750bL4lF6cxaegxiZx5cQFtKZTsE9P6NMYU8v0VhICwmyHYrLjkXkVdfYGzB87C1Hj0L8suUtn9jY7hFTdQQ+AHqZx6nIY6YwKo0E/icn16CrXSZvDYsTldaYXBkJxyFLlorCLZTbKuGV3I966FMInP7y76CTARDKDf33+EloabPjo7bvKc5CruOtQC771o3O4MDS15jS6V/vH8M5AGJ+4t2fNorg7m+xY6qlnq/nhywNIphQ8cmdHSfbX3erCr06P6o26q+hDAH8ohrZG26YVsysxyBLaPXYMjm2839tCE1MJTi3MYtFFREQF27OjFmaTjH5fuKii6+roNNIZFb1FBGAsHC1D/fV1MQNje/DDqxH8/e/cCUkS0FIJqHNT0GJT0OZnoCVmoc3PQkvMYnI8CAlh1CthpH3XgOTyzYUBAWGrheRsguT0QLj0S/37JghTdZw0eN1WqJqG4NQcWrOjKPmRrjKledmtJsQS6XWNEARCMTS7rRWLwl6P2/Z78PrZcfT7IkWvRXvqFR+mZpL4vz51IF8Ub6abexpgqzHg1f7RVYuuUDSB775wBXvba3Hf4bXXnF1f1zW9KUXXTDyFnxy7iiP7PPnX+Ublkkt9gemqKbpUTUMgFMddh7yVPhQAQEezE8fOjEFVtZJNiw1NJSoSwFKNWHQREVHBDLKE/bvqcdoXKWqq0cXhKASA3UX251rOwc56/PrcOIYmZtHhdUKYLJBNFqB2aePTp/7zNCYMc/i/f/s2CCGgqQq0+ZhelGULs1yhpsbC0KaDyAwvXW8mLM6lxZhLvxTmzTuBa1kQG7+k6LKVL0hD0/RYcbuluCmagXB8y51w9e6sg9NmwvFzE0UVXZdHonjxZAD3HW5bki65WYwGGbftb8Yr7wQQS6SX/X2pqobvPHMekgA++9F9BRXSbY12WMwyrgSmcfuB8vQbW+i5N4eRTit4aIV1ZuvRXG+FrcYAX2Aa7z9UHU2Sw9EEkmmlahqHd7Q48MJJP8Yi8ZIUu7FEGnPJTMnj4rcqFl1ERFSUvi43Tl4OwR+KFxxzfHFoCu0eR0nWVe3rqIcAcOZqJL/4eznJlIILQ1P44M2t+eJQSDKE1QVYVz8p1lIJqDPB7NcEtGn9UgmcR+bya4u2FWY7hL0OwmTVCzCTNX9dmHPXrYDZdv12kxUwmIqeUtRcn4uNvz5iNx1LwWKWV11jtxG5BskrncSvJJHMIDw9j7uq5AS3ULIk4UhvE14+FVi2EfVy0hkF//vnF+F21uBj799Y6MNG3dXnxQtv+3H83DjuO7xjyf3PnxjBZf80PvvRXrhdNQXtU5IEOltcGPBHS324S2iahhMXg7il15PvTVcKQgh0tbrgGy19L6r1Ggnq/48rHRefk3s/vTo2U5KiKx8Xz6ILAIsuIiIqUl+XGwDQ7wsXdLKQSivwjU7nEw83ymk1YZfXgbNXJ/HQ+1Ze73H+2iQyiopD3e6if4YwWSA37MyHgCyk5ZpH54qx6Qmoc1F9/dlMEFpyDlpqbklk/hKSrBdhNXYIay2EtfZ6IIi1FpK1Nn9dyPqfa7NJhttpxtjk9WCLaCwJV5lGuQB9pAuAHhtfRADkaLYwbKuSqVzFuH1/M375lh9vXQoVNCryk9evYXxyDn/46CHUmCp7atXucWCnx4FX+8dw7y1tiwp7fzCG/+9XPrxndyPuKHLEqqc112C4sEJ0vUYjcwhPz+O3PrSn5PvuanGi3xcp+3MoVCAUgwDQUiX/Rzz1VljMMq6NzeKuvo3vj3Hxi7HoIiKiotTazdjpceC0L1JQWIAvMI2MomFvEeu51nKgw41nfn0N8fn0iqNnp31hWMwydu/Y+JTGhYTBDLm+DXL96kWkpmagpRJAMp4vxK5fxoHc9/OzUOeiUMcvIzMXBVRl6c+scUDYaiGsdfivFhXRiRqkzgch2eugzk6izla+KXy5ka7ZRHFhGoFQNrlwi8TFL7Sr2QFPnQXHz42vWXQNT8zi58eH8b4DzTjQUXyBXw53HfLiyecvY3gilk/+yygq/umZ87CaDfid+/cUPcra3aY3GPaNzuBgZ/meZ78vDAA4vNcDZDIl3Xdu2ufV0RkcKONzKNRIKIamOsuShtSVIgmBnR4HBsdKMxoYnNI/HNqs8JVqx6KLiIiK1telFz2FTDm7MByFJERJi5+DnW785PVrOH9tCu/du3TdjappOO2LYH+Hu2IhDkIyQNQ4gJrC4641TYOWjEGLT+nR+XP6pTY3BTV72aGGYVLiSB57CwDwaQAZGBB/qgVSXQuk2hZIda2Q61ognE0bTmXM/X5ni2yQHAjFYTJIRTXqrRZCCNy2X+97NTWbRN0KISWKquJ//fwibDUGPHpvzyYf5cpu3efB914YwK/6R/GpZn3E6EfHBjESjOEPPn4QTmvxSZcdXieEAK74p8tbdA1E0NZoR2OdBaFQaZP0cs9hIDBdFUWXPxireH+uG3V4nXj+xMiihNT1CkYTqLWbqqaorDQWXUREVLS+br3oOXs1gtv2rz5N6eLwFHY2O2Axl+5PTkeLA1azAWeuRpYtuoYnZjEdS+FQV+VPrIohhLheqLnbl93mzVMB/NtzF/D4p/fDJWbxv//jFbzXm0GPNQFl/AoyA8evbywZILma9WKsrjV72aKHgsiF/T4c2RP0WJENkgPhGFoabFXVE6kYt+334EfHBvHG+Qncf+vyv4tfnPBjaHwW/8cjB4oOGSknW40Rh/c04o1zE3j0g90YnojhZ8eHcGefFzf3NK5rnxazATua7GVd1zU3n8YV/zQ+ctvy/94bZTEb0Npgh68KmiQn0wqCU4k13z83W4fXCUXVMBKMbTgEJzSV4HquBVh0ERFR0Tq8TjisRvT7Vi+6kikFg6Mz+PCRpQv6N0KWJOzrqMfZq8unKL5zJQwB4OAWK7oK0eK2QoWEsXkTarwdeD3hx45d3Th0RD9R1dLzUKNjUKdGoU4FoEyNQglfQ+bqCQCavhMh6+mLLg+Ey6MXZi6PXozZavONpQG9d5XJIOlruooQCMdxYI1eUdXMU2dFh9eJ4+fGly26Jqbm8MNXr+LmngYc3rO+Qqac7urz4vj5Cfz63Dh+fnwYbmcNfnuDo3E9rbV49cwoMopalhHks4OTUDUtv260HLrbXHjj/HjFmySPhuPQgKpJLszJhWkMjs1suOgKRhNr9ot7N2HRRURERZOEwMFON04PhFft6XIlEIWiaujdWbr1XDkHO+rx1sUgAuH4khOX074Iulpd65pGVe2aF8TG1zn09DmX/frzFMYayI0dkBsXh4xomZRejEVH8wWZOj0B1X8GUBasnZFN1yPxs0XZfmsY6ZgFmtZV0FqgWCKN6VgKrQ3670VTFSCTgqakgUwSwmiBqKmuk83l3Lbfg+/+8goC4fii3k6apuFffn4RBlngkx8ufn3UZtizsw4Nrho8+fxlqKqG//Oxmzc82tzd5sILJ/3wh2LY1Vz6dYT9vghsNQZ0tZQvcr+rxYmXTwUwusz7xmbyB2MAgLam6lrzWO80w2k14toG13Ul0wqisRRHuhZg0UVEROvS1+XG62fH4RudRk/b8uu1LgxNQZYEelpLG2YBIL8m4+zVyUUnT1OzSQyNz+Ljd1c2urtcnFYjbDUGvVdXthCos6+dXigMpmUTGTVNhRaf0guw6fHs5QTU6Cgyw+8AqoLfNQAYB2L/658hZXuVBR12zMfmoGVSgJLWi6pMClBSSM/P4xu1CdjOapjtzwDa0nAQydUMubkHsqcHcvNuCJen6oqXI70e/McLAzh+bhwfv7srf/ur/WO4OBzFp+/fs+J6r0qThMCdfV48/eogjh7ZgT0lCLLpadOLoSv+6ZIXXaqm4czVCA52ukvWmHc5uUbPvsB0RYuukVAMJqNUdSETQgjs8joxOL6x9XS5uHgmF17HoouIiNblQEc9JCHQ74usWHRdHIqio8VZlh5SdQ4z2hptOHM1smj615mrEQDAoa6Gkv/MaiCEQLPbivFIHNGYfuLrKqDoWnl/EoTdDcnuBlr3LbpPU1VosQj+88evwZaaxIf2mqFOT0CZHMF8RIEqDIDBpPcck416XzKDCdHpNC5F53Db3jaYrFbAYISQTflLdX4GyvgVpK+dRPrSq/px1Dgge7ohN++G3NwDqWEnhFzZdVIumwn7dtXhjfMT+Nj7OyGEwNRsEv/x4gD2ttdWTZPdldx3yw6YjTLueU9rSfZX76xBvdOMAf80PrRMD7CNGBybwexcuqxTCwGgqc4Cu8UIX2AGd99Umn+X9QiE4mhtsFflmscOrxNnfBEkkpl1j46GsnHxnjprKQ9tS2PRRURE62KtMaKnzYXTA5FFowA5iWQGQ+Oz+I3bl/a6KpUDHW788u0RzKcy+f5IpwfCcDvNWzKqvFBetw39vgimY3qMu8tWnmmUQpIgnI2YdvbgndEZPHjH7fn7GhsdK6bLHXvuEo4rEzh6912rjl5pmgo1Og5l/DKUiStQxgeQGTql3ykbIDd2Xh8N83RXZEribfs9+PYzFzAQmEZ3qwtPPn8JGUXFpz+yt+pG5m5krTHg6JHShlJ0t7pwxT+97FrKjegfiEAIlD1VUAiBrhYnfKOVC9PQND2o4j27q/ODoQ6vExqAofFZ7F3n1PBgbqSrykbyKolFFxERrVtftxvff8mHyZl51DtrFt13eSQKVdPQ2176qYU5Bzrr8eybw7g4HMVN3Q1IZxScuzaJOw96q/6EeCO8biuO9Y9hNByH2SSXNBlyOQ6LEbNFpBcGQjG0ZQq2IwAAIABJREFUNtrW/B0IIUGua4Fc1wL0fgAAoM5NZwuwK1AmriB1+llA+ykA6MmL7nYAQu9npirQ1AygqQu+V/LXl3yvqYBsWDTqBoMJkI0QhuxtBvP172UjDsGAuy3DCLwRQaq1FQMDU3jk7v3v2k/we9pq8eaFICIz82hwle6Euj+7DnMzUiC7Wl047YsU1PKiHKbjKcQS6aoL0cjp8OptLgbHZzZUdFnNhqpK9aw0Fl1ERLRufV0N+P5LPvT7IvjAzYun6lwcnoJBFvmGpOXQ01YLk1HC2asR3NTdgAtDUaTSKg51V+cnyKXizYZpXBieQm2ZRrkWsluNSCQzBaXWaZqGQDi+bJR/ISSrC1LHYRg7Duv7yyShBAcXFGIDgJD0hEXJAEgyIMl6PzJJ0oum/PcyIKTsNgYISUBTMgtCPVLQMklgfhZqJgVk16fl1qZB09MeP2YBENG/vlEH4NLPMDexsC+a/iVs9du62Aeur4ka8E+XrOiKxpIYmti8dZgLmySXezrjcvyhbIhGlRZdDqsJDa4aDI6tf11XaCrB9Vw3YNFFRETr1uK2osFVs3zRNRRFV4sLpjI2xjQaJPS21+Hs1UkAwGlfGCajhL1lHF2rBl63PsoyOZPEnhI2nV6JI/tpdTyRXnP9WDSWQnw+g9YSnVAKgxmGlr0wtOwtyf4KpWkaoOoF2tmBCfzrT/vRIMfx6dudcCkRqFOjyAy+DS35yvUHGWsg1XrzhZicLcqEoxFCqkyT7lJra7LBbJRxJTBdsh5TZ3z6Osy+TVqH2eF15JskV6ToCsYBoOoaIy+0y+vcUIJhMJrATk/hjeHfDVh0ERHRugkh0NflxrEzY0hnFBgNeoEVn09jeGIWD93ZscYeNu5ApxunfRFMTM6hfyCM/bvq88exXTW6LDDIAhlFWxQXXy72bPT+bAFFVyCsf4q/MGJ9KxJCALIRkI3o3bsLhteC6N3vgeeOxa9pNTGT7YsW0KP4o2NQAueQufLa9Y1kA4TFpU9blAyArH/lr0sGvVm1bAAk44Lr2duFlJ9CqS2YSglVBTQFmpq9TdNv0xZcBwBhrYVkr88GpmQvbfUQpuJHImRJQmeLEwP+0q2J6vdF8sE4m6HGpDd6rlSTZH8ohlq7qaqn3nV4HXjrYhAzc6miW28oqorI9Py6R7u3KxZdRES0IX1dDXjxZAAXh6M4mF0Ef3k4Cg3YlBGng516881n3xxGZCaJB99X/kKv0iRJwFNvRSAUR+0GkgsLlTs5LKRBciCkf4rfso2CTAyyhL/6/G3L3idZnJAsTsC7Z9HtWmou2w9tFEp0FNr8rN4PTcno69CUDKBmoKXm8rdpyoLblTSgKICa1qc5CgmQJEBkp1FKhgVTJ/VLkb0vtw0kGdBUqKNjyMxN5adL5pmsi4uwhUWZvR7CVqf/nBv0tLnwk9evbSjdLiejqDh3bRK37dvclgFdrS68fnZ81T6D5eIPxqp6lAsAOrNNkq+NzRQ9AhmZSUJRNfbougGLLiIi2pC97bUwGST0D0TyRdeF4SkYDRI6y9jkNKepzoqmOgt+9c4oAFRkulAleN22TSu6ctMLCwnTCITicNpM27IxdTGEyapH4Hu6sdHxjFIkBWqqAm0uCjU2CS0WWXAZgRafRHpiAEjGFz9ICH2qZOMuyA0dkBt3QXK3o7vNBU3T10Tt76jf0HFdHoliPqVs2tTCnO4WF146GUAgHMeOTSyAMoqK0Ugc+zb471Zu7R4HBIDBsdmifze5uPgmrulahEUXERFtiMkoo3dnHU77wnhM64EQAheHouhudcFo2Jx1LAc73Hhhyo9dzY5NKUKqgbdeX9e1GdMLHdbcSFdqzW0D4diWn1pYbUoxAiQk+Xo/NvQsu42WTkKNR6DFJvVibDYMJTIMZeQMMpez0yWFjJ11LfiEzYy5MxEo9vdCqm/Tp0GuQ78vAoMsoXedKXnr1dWqj+T4AtObWnRNTCWQUTTsqNIQjRyL2QBvgw2D61jXxbj45bHoIiKiDevrbsBpXwRjkTk4rEb4QzH85vs3J4kM0KPjXzjp3/aphQt5G/SiazOKTFuBI11qNrmw2psG0/KE0Qy5tgWoXfz70zQNWnwKSmgQamgQSvgabjJfgWV8AHM//CkgGSC5d0Bu7IDcsAtSU4ceICLJC/ahAkoayOTSIdPQlBSCvvP4QIsMeeI80koKyKQBJY1Y0AEFDghnE4TZXvKph421FjisRvgC00tCgMrJH8wmF1b59EIA6Gh24MzVSNEjraGpBAyyhFrHu+MDsEKx6CIiog3ry04r7PdF0ODS+3X1tm/eJ9f7O+rx0dt34gM3vXtO9m/ubsTH3t+JnrbyT+E0yBIsZsOaa7rC0/NIpdWqjcKm9RFCQNjrIdnrgY5bAABPPXcRF84P4CsPNgKRISihQaSv/Brp8y/qD5JNEKaabDR/Wk+CXMZ/B4A4kPj54tvnF35jtEByNkFyNkJyNkE4m/LfC5t7XcmQepNkFwZG15/Qtx7+UAyyJPIJpNWso8WJ186OF92TbWJqDk11FkjbvH1CsVh0ERHRhrldNWhrtKHfF0ZLgx4pvcu7eXHBBlnCx+/u2rSfVw3MJhkP3LFr036ew2JEbI2RrkBoeyQX0tq622rx0ikrxu37sLP7VgD6aJY2HYQSHoQSugakk9mG06ZsEqQJwmDMN5/uvzaDl/qD+N0HDqKu1qFvYzBCyEbUOQyIXLsGdSaof82GoEz6kRk6pSc15kgyhKNBL8IcejEm7HV6yAgEIABAQCy4nrs87Azi2NA4ZgfMsNTkVt4JPYQkeymEpO9LiOzl9eti4e2Sfp8wWQCDecWRIX8whma3dc1+d9WgIx+mMVtU0RWKJhiisQwWXUREVBJ9XQ147s1hhKfn0bPDtSVOKqhwdqtxzemF+eRCFl3bXk+uSXJgGjub9Q9YhJAgapsh1TbD2H37mvt4+e1TmHLWomHPzUvuMzU6YMDS0XJNVaHFJ/PFmJYrymZCSI8PAOlEwc/hIICDDgAvvoDCH1UA2QhR44CwOPTLGgeExQlR40BjOIgudz0y41cgZbeByVqVTbXbGu2QJYHBsRkcLjD+XdM0hKLz6N1Z3UEhlcCii4iISqKvy42fHR9CeHoeH9zENRK0OewWI6Kx5KrbjIbjcDvNG44Rp+rndtWg1m7CFX8U997SVvTj51MZXBop/rFCkvSRLUcD0Lpv0X2apgHJONRcPH4+Il8DtNzl9dtSaQX/7/dO4bb9Hnzgptbr92ua3hct/6V/ry24rvdA07K90rL3qSq0VALa/Ay0+VloiVlo87NQp8ehJWaBTBIPygCiQOLHTy94UjJEjR3C2QjZ3Q6pYSdk905I9a0QcuV6eRkNEnY02YsK05iJp5BMK0wuXAbfFYmIqCS6Wp2w1RgQn89g7yYnkVH55QJSVuMPxdHK9VzvCkIIdLfVrrvB8PlrU8goWkmj4oUQQI0dck1hr0ELAMU9jZOTMu5rXj7RsZQuXwviW//5Bj73oR3oaZAXFWba/AzU6Qmkr7wO5NbFCRlSXQukhna9CGvYCdndvq6m1uvV4XXi1+fGoWpaQWu0mFy4MhZdRERUErIkoa/LjX5fBO0ennhvNw6LadUgjYyiYnwynm9WTdtfT6sLb10MYnJmHvXOmqIe2+8Lw2KWNyUIZjXdLS4cOzMGRVUhryOQoxiByRSiqg2err0wrPDvpWkqtJkQlMgQ1PAwlMjQ4sh+AMLpgdzQDsm9M38pWV36SJ+ipz/mEyIzKSCTygaapLK3Zy8zaf26puoJkRZndsqjE8LiBIw16PA68dKpAMYjcwVNGw5me3R5ONK1BIsuIiIqmd++bzceuCNV9pMX2nx2qxGpjIpkWoHZKC+5P5jtP9TayPVc7xbdbdfXdR0poujSNA39vgj276qv+NrPrlYnXjjpRyAUR7unvOE/I6E4rGYD6laJUhdCgnB5ILk8QOcRANnI/rko1MgQlPAw1PAQlNA1ZK6euP5A2QgoGWTnUZaGbMBBkx1/5JSgvXgcicaG7No0JyRLbs2aU//ZkgQhZMxOjKJOiqPeOA81kdHbBggJkORs2IhclevXNgOLLiIiKhm7xQi7pXJrEKh8cr/X2FwaZtfSoisQ1kM0Whs4yvlusaPJDpNRwhX/NI70egp+3PBEDNFYqqRTC9erKxsI4gtMl73o8odiaGu0FV10CCEgbHWQbHUwtN+Uv11LxqFE9CJMnYvq678MJgjZlE2NNGfTIE3Z27P3G/TvcymSEBK0+Zi+Fi0xCy0xk7+uJmYQO38N9rlZKOOT0OZmAGXlJunvA/C+WmD+u0+t8oSkfAEmuTyQvXsge/fC4N0DUeDU0K2IRRcRERGtyZEruhJpuF1LRzUCoRiEwJboP0SlYZAldHqdGPAXt66r3xcGABzscpfjsIrS4KqB02bCQGAGH3xP+X6OpmnwB2O440BzyfYpzDYYWnqBlt6N78teD9iXnxr8YuAkMoqKP//twwAALZ3MFmb6WjRNUfQYf03Bj171wSQD97+3Tb8te7umqtnr1y81JQM1Moz0hVeQPvsLAIBU16YXYS17IDfvgWSt7PTTUmLRRURERGuyW/Wiazax/KfcgVAcTXVWmJaZekjbV3dbLX726yHMpzKoMRV2Wtnvi6DD64DLZirz0a1Nb5LshG90fYEghYpMz2M+pWzJxuEdXgdeeNuPjKLCIEsQRjOEsRFwNi7Z9lc/zeCWPY0w9e4teP+akoESGoQydhHK2CWkLx9D+vwLAACp1psfCZO9eyDZtm5IE4suIiIiWpPDqp8gz64QpuEPx9HG/lzvOt2tLqiahqujM9i3a+0QlZm5FK6OzuChOzs24egK093mwqkrYczMpeC0lqcQ9Gd72LU1bcWiy4mMosEfimFXs3PF7ebmM4gl0kXHxQvZAENzDwzNPcDND0JTM1BD15AZu6QXYQPHkb7wsr6t0wODdw9k7x4Ydt4EYd467zksuoiIiGhNC9d03SiVVhCcmsOtvYU1UKXto7vVCQFgwD9dUNF19moEGvS+ftWiq+X6uq6be5aO3pTCSLbdQusW/GCiw6sXWoNjs6sWXaFsXHzTBuPihWSA7OmG7OkGbvooNFWBGhmGMnYRmdFLSA++hfSlX8HQeQSW+76woZ+1mVh0ERER0ZqsNQYIAcwmlhZdY5E5aBrYo+tdyFpjREujDVcK7NfV74vAaTNhZ3N5QyuKsavZAVkS8AVmylZ0BUIxNLhqtmTj8AZXDewWIwbHZlZtfF+uHl1CkiE3dkBu7ICp7yPQVBXqlF+Ptd9CmOlLREREa5KEgN1iRGyZoisQ3rqf4tPG9bS6cHV0Gqq6ely5oqo4e3USfZ3ughrtbhaTUUa7x77uRs+FGAnGsGMLTi0E9HVvu7wODI7NrLpdcGoOQPkbIwtJguxuh2StLevPKTUWXURERFQQu8WI2NzSII1AKA6DLIpey0HbQ3ebC4mkkm8bsBJfYAZzyUxVTS3M6WpxYXB8Boqqlnzf6YyCicnElh4J7mh2YjQcRzKlrLhNKJqA02rckqN5m4FFFxERERXEseJIVxzN9baKN7qlyuhu00ccBvzRVbc77QtDlgT2d6y99muzdbW6kEqr8AdXLxzXYzQ8B1XTtuxIFwB0tDihacDQxOyK2wSnEmiqY8uIlfDdkYiIiApit5qWTS8MZJu+0rtTo6sGLptpzXVd/b4Idu+orcqRkK5WfX3QQBmmGPqzIRpb+f/I9TCNlacYBqOJsk8t3MpYdBEREVFBHFbjkiCNRDKDyEwSrVv4hJI2RgiB7jbXqk2SI9PzCITiONhZfVMLAcDtrIHLbipLvy5/KAajQdrS029dNhPqneYVi650RsXUTHJLP8dyY9FFREREBdHXdKWhadcDE3LreFoYovGu1t3qQnh6HlOzyWXv7/eFAQCHuquz6BJCoLvVVZYwDX8whpYGG2Rpa592dzQ7Vyy6wtMJaNh4XPx2trV/+0RERLRpHBYjVE1DIpnJ3xbI9R/awiEBtHHdbXqvq5Wm5532RdBYW4Pm+upd89PV4kIoOo/p+NKwmI3wh+JbemphTkeLE6Ho/LLrOoNT2bh4jnStiEUXERERFcRu1RskL5xiGAjHYTJKaHDVVOqwqArs9DhgNEi4skyYRiqt4OLQFPq6GiCqKCr+Rrt36IEgpwfCJdvnTDyF6XgKO7bBhxId2d5q15YZ7cr16OL0wpWx6CIiIqKC2C0mAEBsQZhGIBRHa4Otqvou0eYzyBI6vM5lp+ddHJ5CKqPiUBVGxS/U4XWgvcmO594chqqt3nOsULkQjdYtnFyYs7N55TCN4FQCNSYZDotxsw9ry2DRRURERAVx5Ea65haPdLU2bP0TStq4njYXhidiSKYX93I67YvAZJSwp726m9kKIXD01naMRebQ74uUZJ/+kL7mcTuMdFlrDGiut2JwbGlsfCiaQFOtpapHMiutoKJrcHAQjz76KI4ePYpHH30U165dW7LNE088gY9+9KN48MEH8bGPfQyvvvpqqY+ViIiIKij3KfZsQl/zMjOXwkw8xeRCAqCHaSiqhsHR6yMhmqahfyCCfTvrYTTIFTy6wrx3bxPqnWY898ZwSfbnD8bgtBrhtJlKsr9K6/A6MDg2syhMB9BHuriea3UFFV1f+cpX8Nhjj+G5557DY489hi9/+ctLtunr68MPfvAD/OQnP8Ff/dVf4Ytf/CLm5+dLfsBERERUGbk1XbmF9KPZT/FZdBGgNxgGsKhf12g4jsjMPPqqNLXwRgZZwocO78ClkeiqPakK5Q/F0LYNphbmdHidmI6nFqVUqqqG8HSCyYVrWLPoikQiOH/+PB544AEAwAMPPIDz589jcnJy0XZ33XUXLBb9H3vPnj3QNA3R6OqdyYmIiGjrMBtlGGQpv6YrFxfP6YUE6C0FWhpsi/p15abp9VVpf67lvP9QCyxmA57d4GiXqmoIhONo2wZTC3OuN0m+PsVwajaJjKIxRGMNa7YEHxsbg8fjgSzrQ8KyLKOpqQljY2Oor69f9jFPP/002tvb0dzcXNTBuN3V96JsbHRU+hBom+NrjMqJry8qNZfdhEx2ZlEkloLdYkRPh5trOQgAcKCrAa/1j8LttkOSBM4PR9HR4sSersai91XJ96/fuGMXfvjyABRJQrN7fSO5gVAM6YyK3s6GbfNe7Kq1QpYEgjPz+ec0GtVntvXscm+p57nZx7pm0VWsN998E9/85jfxz//8z0U/NhKJQVVLkxZTCo2NDoRCSxcLEpUKX2NUTnx9UTlYzQaEJucAAAMjU2hxWxEOxyp8VFQt2txWxBNpnL44jjqHGRcGJ/GR29qLfi+q9PvXHfs8ePoVH7733EX8tw/tXtc++i8GAQC1FsO2ei9ubbTh7EA4/5yuXNNHM01C2zLPsxyvL0kSqw4grTm90Ov1YmJiAoqiJ9EoioJgMAiv17tk21OnTuGP//iP8cQTT6Czs3MDh01ERETVyG4xYjaRgqZpelz8Npo6RRvXk2uS7J/GucFJqJqGQ10NFT6q4tU5zLhtnwev9o8u2wy4ECPBGIQAvO7qbQi9Hh1eJ66Nz+Zj9YPRBGRJoN7BXn2rWbPocrvd6O3txTPPPAMAeOaZZ9Db27tkamF/fz+++MUv4h/+4R+wf//+8hwtERERVZTDakRsLo3I9DwSyQxDNGiRpjoLHFYjBgLTOD0Qgd1iRGeLs9KHtS5Hj7QjlVbx0qnAuh7vD8XQXG+FyVj9qY3F6PA6kUhmEJzSGyKHphJoqLVAkjjFeDUFpRd+9atfxZNPPomjR4/iySefxNe+9jUAwOc+9zmcOXMGAPC1r30N8/Pz+PKXv4yHH34YDz/8MC5dulS+IyciIqJN57CYEEukMTSuJ7u1NrDoouuEEOhudeHySBRnrkZwoLN+y56MtzXZcaCzHi+87Uc6o6z9gBv4Q7FtORJ8PUxDfw8IRhPwMERjTQWt6erq6sL3v//9Jbf/0z/9U/76U089VbqjIiIioqpktxoRn8/gajYWfDueVNLG9LTV4tSVMACgr2vrpBYu5/4j7fh/vvcOfn1uAu8/1FLw4+ZTGYSi87jz4NLlOFtdS4MVJoOEwdEZ3LbPg+BUAj1t1d34uhoUNNJFREREBOhrugDg3NUIXHZT/nuinO7sui4hgAMdW7vo6t1Zh3aPHc+9OZxfw1SIQLaH3Xbq0ZUjSxLamx0YHJ/BbCKN+ZTCHl0FYNFFREREBXNkGySfH4xwaiEta6fHAYMsobvVteWLciEE7j/SjrHIHPoHIgU/biSkJ3pupx5dC3U0OzE8EcN4RE8ybeT0wjWx6CIiIqKC5U6iE0mFTZFpWUaDhE99eDc+fndXpQ+lJA7vbYLbacazbxbeLDkQjKPGJMPt2p6Jfh0tDqQzKt4Z0KeRcqRrbSy6iIiIqGAOqyl/ncmFtJK7DrVg947tsc7HIEv40OEduDwSxdXRmYIeMxKKobXRBmmbNg3PhWm8eWECAkBj7fYsLkuJRRcREREVbOF0MRZd9G5x16EWWMyGgka79B52MezYplMLAX1ky1ZjwORMEnVOM4yG7RWLXw4suoiIiKhgC4uuFjeLLnp3sJgN+MDNLXj7UhDBaGLVbadmk4jPZ7Z1sqcQAruaHQA4tbBQLLqIiIioYEaDhBqTjKZ6KyzmgjrPEG0L992yA5IQ+MWbI6tu58+GaOzYhsmFC3Vkm143sugqCIsuIiIiKorLZsqv6SB6t6hzmHHbfg9ePTOKWCK94nb+XFz8Np9+29Gsvwc0MbmwICy6iIiIqCiff2g/PvPQ/kofBtGmu/9IO1JpFS+d9K+4jT8YQ73TDGvN1o7LX0vPjlo01Vqwp72u0oeyJbDoIiIioqJ0eJ1oYVw8vQu1NtpxsNONF972I51Rlt3GH4pt2/5cC9ktRvz1/7gd3a2uSh/KlsCii4iIiIioQPcf2YGZuTRePzu+5L6MomIsMveuKLqoOCy6iIiIiIgKtHdnHXZ6HHjuzRGomrbovvHIHBRVQ1vT9l7PRcVj0UVEREREVCAhBI7eugPjk3M4PRBedN9INrmQI110IxZdRERERERFeO/eJridZjz3xuJmyf5gDLIk0FxvrdCRUbVi0UVEREREVARZkvCh97bjsn8avtHp/O3+UBwtDTYYZJ5i02J8RRARERERFemuPi+sZsOi0S49uZDruWgpFl1EREREREWymA34wM2tePtyCMGpOcQSaUzNJtHWxPVctBSLLiIiIiKidbjvcBskIfD8iREEGKJBqzBU+gCIiIiIiLaiWrsZt+9vxrH+MdgtRgAsumh5HOkiIiIiIlqno0d2IJVR8ewbw7BbjKi1myp9SFSFWHQREREREa1Ta6MdfV1upDIq2hptEEJU+pCoCrHoIiIiIiLagPuPtAPQCzCi5XBNFxERERHRBuxpr8VvfbAbh7rdlT4UqlIsuoiIiIiINkAIgftvba/0YVAV4/RCIiIiIiKiMmLRRUREREREVEYsuoiIiIiIiMqIRRcREREREVEZsegiIiIiIiIqIxZdREREREREZcSii4iIiIiIqIxYdBEREREREZURiy4iIiIiIqIyYtFFRERERERURiy6iIiIiIiIyohFFxERERERURmx6CIiIiIiIiojFl1ERERERERlxKKLiIiIiIiojFh0ERERERERlRGLLiIiIiIiojIyVPoAFpIkUelDWKIaj4m2F77GqJz4+qJy4uuLyomvLyqnUr++1tqf0DRNK+lPJCIiIiIiojxOLyQiIiIiIiojFl1ERERERERlxKKLiIiIiIiojFh0ERERERERlRGLLiIiIiIiojJi0UVERERERFRGLLqIiIiIiIjKiEUXERERERFRGbHoIiIiIiIiKiMWXURERERERGXEomsFg4ODePTRR3H06FE8+uijuHbtWqUPibawxx9/HPfccw/27NmDy5cv52/n64xKYWpqCp/73Odw9OhRPPjgg/j93/99TE5OAgDeeecdPPTQQzh69Cg+85nPIBKJVPhoaSv6whe+gIceegiPPPIIHnvsMVy4cAEA38OotP7xH/9x0d9Jvn9RKdxzzz24//778fDDD+Phhx/Gq6++CqACry+NlvWpT31Ke/rppzVN07Snn35a+9SnPlXhI6Kt7MSJE9ro6Kj2wQ9+ULt06VL+dr7OqBSmpqa048eP57//67/+a+1P//RPNUVRtPvuu087ceKEpmma9sQTT2hf+tKXKnWYtIXNzMzkr//iF7/QHnnkEU3T+B5GpXP27Fnts5/9bP7vJN+/qFRuPPfSNK0iry+OdC0jEong/PnzeOCBBwAADzzwAM6fP5//5JioWIcPH4bX6110G19nVCq1tbW49dZb89/fdNNNGB0dxdmzZ2E2m3H48GEAwCc+8Qk8++yzlTpM2sIcDkf+eiwWgxCC72FUMqlUCl//+tfx1a9+NX8b37+onCrx+jKUde9b1NjYGDweD2RZBgDIsoympiaMjY2hvr6+wkdH2wVfZ1QOqqriu9/9Lu655x6MjY2hpaUlf199fT1UVUU0GkVtbW0Fj5K2oj/7sz/Da6+9Bk3T8O1vf5vvYVQy3/zmN/HQQw+hra0tfxvfv6iU/uiP/giapuGWW27BH/7hH1bk9cWRLiKibeQv/uIvYLVa8clPfrLSh0LbzF/+5V/i5Zdfxhe/+EX8zd/8TaUPh7aJU6dO4ezZs3jssccqfSi0Tf3bv/0bfvzjH+Opp56Cpmn4+te/XpHjYNG1DK/Xi4mJCSiKAgBQFAXBYHDJ9DCijeDrjErt8ccfx9DQEP7+7/8ekiTB6/VidHQ0f//k5CQkSeKnxLQhjzzyCN544w00NzfzPYw27MSJE/D5fLj33ntxzz33YHx8HJ/97GcxNDTE9y8qidx7kslkwmOPPYZ4Q0OeAAABqklEQVSTJ09W5O8ji65luN1u9Pb24plnngEAPPPMM+jt7eV0CSopvs6olP7u7/4OZ8+exRNPPAGTyQQAOHDgAObn5/HWW28BAL73ve/h/vvvr+Rh0hYUj8cxNjaW//7FF1+Ey+XiexiVxOc//3kcO3YML774Il588UU0NzfjO9/5Dn7v936P71+0YXNzc5idnQUAaJqGn/3sZ+jt7a3I30ehaZpW1p+wRfl8PnzpS1/CzMwMnE4nHn/8cXR2dlb6sGiL+sY3voHnn38e4XAYdXV1qK2txU9/+lO+zqgkrly5ggceeAC7du1CTU0NAKCtrQ1PPPEETp48ia985StIJpNobW3F3/7t36KhoaHCR0xbSTgcxhe+8AUkEglIkgSXy4U/+ZM/wf79+/keRiV3zz334Fvf+hZ2797N9y/asJGREfzBH/wBFEWBqqro6urCn//5n6OpqWnTX18suoiIiIiIiMqI0wuJiIiIiIjKiEUXERERERFRGbHoIiIiIiIiKiMWXURERERERGXEoouIiIiIiKiMWHQRERERERGVEYsuIiIiIiKiMvr/AQA6zH3vLFQMAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 1058.4x595.44 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iz76m5LIP9Tx"
      },
      "source": [
        ""
      ],
      "id": "iz76m5LIP9Tx",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "AxxpEAiOP-vy"
      },
      "source": [
        "## Resultados em dados de teste"
      ],
      "id": "AxxpEAiOP-vy"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7UgGSNKcLCDw"
      },
      "source": [
        "# RANDOM FOREST\n",
        "\n",
        "from sklearn.ensemble import RandomForestClassifier\n",
        "\n",
        "rf = RandomForestClassifier(max_depth=3, n_estimators=10)\n",
        "rf.fit(X_train, y_train)\n",
        "\n",
        "y_pred_rf = rf.predict_proba(X_test)[:, 1]\n",
        "fpr_rf, tpr_rf, thresholds_rf = roc_curve(y_test, y_pred_rf)\n",
        "auc_rf = auc(fpr_rf, tpr_rf)"
      ],
      "id": "7UgGSNKcLCDw",
      "execution_count": 227,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 587
        },
        "id": "an93OngvLBRG",
        "outputId": "49f71ab3-9709-46d7-9c1e-6cbc09f2d92c"
      },
      "source": [
        "from sklearn.metrics import roc_curve\n",
        "from sklearn.metrics import auc\n",
        "\n",
        "y_pred_keras = model.predict_classes(X_test)\n",
        "# y_pred_keras = model.predict(X_test)\n",
        "# y_pred_keras = np.array([1 if p > 0.95 else 0 for p in y_pred_keras[:,1].tolist()])\n",
        "fpr_keras, tpr_keras, thresholds_keras = roc_curve(y_test_k[:,1], y_pred_keras)\n",
        "auc_keras = auc(fpr_keras, tpr_keras)\n",
        "\n",
        "plt.figure(1)\n",
        "plt.plot([0, 1], [0, 1], 'k--')\n",
        "plt.plot(fpr_keras, tpr_keras, label='Keras (area = {:.3f})'.format(auc_keras))\n",
        "plt.plot(fpr_rf, tpr_rf, label='RF (area = {:.3f})'.format(auc_rf))\n",
        "plt.xlabel('False positive rate')\n",
        "plt.ylabel('True positive rate')\n",
        "plt.title('ROC curve')\n",
        "plt.legend(loc='best')\n",
        "plt.show()"
      ],
      "id": "an93OngvLBRG",
      "execution_count": 228,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/keras/engine/sequential.py:450: UserWarning: `model.predict_classes()` is deprecated and will be removed after 2021-01-01. Please use instead:* `np.argmax(model.predict(x), axis=-1)`,   if your model does multi-class classification   (e.g. if it uses a `softmax` last-layer activation).* `(model.predict(x) > 0.5).astype(\"int32\")`,   if your model does binary classification   (e.g. if it uses a `sigmoid` last-layer activation).\n",
            "  warnings.warn('`model.predict_classes()` is deprecated and '\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1058.4x595.44 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "17570055",
        "outputId": "f3862f2f-95f0-413c-ba9c-7f729803a0a2"
      },
      "source": [
        "from sklearn.metrics import classification_report\n",
        "from sklearn.metrics import confusion_matrix\n",
        "from sklearn.metrics import precision_score, recall_score, f1_score, accuracy_score, roc_auc_score\n",
        "\n",
        "y_pred = model.predict_classes(X_test)\n",
        "# y_pred = model.predict(X_test)\n",
        "# y_pred = np.array([1 if p > 0.95 else 0 for p in y_pred[:,1].tolist()])\n",
        "\n",
        "#print(classification_report(y_test, y_pred))\n",
        "print(confusion_matrix(y_test, y_pred))\n",
        "\n",
        "print(\"Precision:\", precision_score(y_test, y_pred))\n",
        "print(\"Recall:\", recall_score(y_test_k[:,1], y_pred))\n",
        "print(\"Acuracia:\", accuracy_score(y_test_k[:,1], y_pred))\n",
        "print(\"AUC:\", roc_auc_score(y_test_k[:,1], y_pred))"
      ],
      "id": "17570055",
      "execution_count": 225,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/keras/engine/sequential.py:450: UserWarning: `model.predict_classes()` is deprecated and will be removed after 2021-01-01. Please use instead:* `np.argmax(model.predict(x), axis=-1)`,   if your model does multi-class classification   (e.g. if it uses a `softmax` last-layer activation).* `(model.predict(x) > 0.5).astype(\"int32\")`,   if your model does binary classification   (e.g. if it uses a `sigmoid` last-layer activation).\n",
            "  warnings.warn('`model.predict_classes()` is deprecated and '\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "         0.0       1.00      0.88      0.94     49331\n",
            "         1.0       0.03      0.70      0.05       239\n",
            "\n",
            "    accuracy                           0.88     49570\n",
            "   macro avg       0.51      0.79      0.49     49570\n",
            "weighted avg       0.99      0.88      0.93     49570\n",
            "\n",
            "[[43378  5953]\n",
            " [   72   167]]\n",
            "Precision: 0.027287581699346405\n",
            "Recall: 0.698744769874477\n",
            "Acuracia: 0.8784547105103894\n",
            "AUC: 0.7890350716859361\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "719b8ce0"
      },
      "source": [
        ""
      ],
      "id": "719b8ce0",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Eujde_mlQmXP"
      },
      "source": [
        "## Salvar modelo"
      ],
      "id": "Eujde_mlQmXP"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "d8733261",
        "outputId": "5d2d6f04-9535-48df-f7ef-046f2bec9f93"
      },
      "source": [
        "# serialize model to JSON\n",
        "model_json = model.to_json()\n",
        "with open(\"model.json\", \"w\") as json_file:\n",
        "    json_file.write(model_json)\n",
        "# serialize weights to HDF5\n",
        "model.save_weights(\"model.h5\")\n",
        "print(\"Saved model to disk\")"
      ],
      "id": "d8733261",
      "execution_count": 241,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Saved model to disk\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TVrnaz-CQqe7"
      },
      "source": [
        ""
      ],
      "id": "TVrnaz-CQqe7",
      "execution_count": null,
      "outputs": []
    }
  ]
}