TP2/7506R_TP2_GRUPO02_ENTREGA_N1.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/kikiymini/7506R-1C2024-GRUPO02/blob/main/TP2/7506R_TP2_GRUPO2_Bayes_naive.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GDDSr-SglU1G"
      },
      "source": [
        "# <center>Organizaci&oacute;n de Datos</center>\n",
        "\n",
        "#### <center>C&aacute;tedra Ing. Rodriguez, Juan Manuel </center>\n",
        "\n",
        "## <center>Trabajo Práctico 2: Críticas cinematográficas</center>\n",
        "\n",
        "### <center> Grupo 2</center>\n",
        "\n",
        "## Integrantes:\n",
        "\n",
        "*   Aramayo Carolina\n",
        "*   Utrera Maximo Damian\n",
        "*   Villalba Ana Daniela\n",
        "*   Fiorilo Roy\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yQzgLrPLld-Z"
      },
      "source": [
        "# Importación de librerias"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "vFcCB64v6gch"
      },
      "outputs": [],
      "source": [
        "# instalaciones\n",
        "%%capture\n",
        "!pip install es_core_news_sm"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "VvM4hVZ9uFpL",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "fdc7623e-e6e4-4710-a78d-a5eeb7478339"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "[nltk_data] Downloading package stopwords to /root/nltk_data...\n",
            "[nltk_data]   Package stopwords is already up-to-date!\n",
            "[nltk_data] Downloading package punkt to /root/nltk_data...\n",
            "[nltk_data]   Package punkt is already up-to-date!\n",
            "[nltk_data] Downloading package omw-1.4 to /root/nltk_data...\n",
            "[nltk_data]   Package omw-1.4 is already up-to-date!\n",
            "[nltk_data] Downloading package wordnet to /root/nltk_data...\n",
            "[nltk_data]   Package wordnet is already up-to-date!\n",
            "[nltk_data] Downloading package averaged_perceptron_tagger to\n",
            "[nltk_data]     /root/nltk_data...\n",
            "[nltk_data]   Package averaged_perceptron_tagger is already up-to-\n",
            "[nltk_data]       date!\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "from matplotlib import pyplot as plt\n",
        "import seaborn as sns\n",
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "# Tokens y preprocesamiento\n",
        "import nltk\n",
        "from nltk import word_tokenize\n",
        "from nltk.corpus import stopwords\n",
        "from nltk import pos_tag\n",
        "from nltk.stem import WordNetLemmatizer\n",
        "from nltk.corpus import wordnet\n",
        "from collections import Counter\n",
        "import spacy\n",
        "from nltk.corpus import stopwords\n",
        "from sklearn.feature_extraction.text import CountVectorizer\n",
        "from sklearn.feature_extraction.text import TfidfVectorizer\n",
        "\n",
        "# nltk\n",
        "nltk.download('stopwords')\n",
        "nltk.download('punkt')\n",
        "nltk.download('omw-1.4')\n",
        "nltk.download('wordnet')\n",
        "nltk.download('averaged_perceptron_tagger')\n",
        "\n",
        "# Modelos\n",
        "import joblib\n",
        "from sklearn.model_selection import cross_val_score, KFold, GridSearchCV, StratifiedKFold\n",
        "from sklearn.metrics import confusion_matrix, classification_report, f1_score, make_scorer\n",
        "\n",
        "from sklearn.naive_bayes import MultinomialNB\n",
        "\n",
        "# Utils\n",
        "from joblib import Parallel, delayed\n",
        "from sklearn.preprocessing import LabelEncoder\n",
        "import warnings\n",
        "warnings.filterwarnings(\"ignore\")\n",
        "\n",
        "RAND_SEED = 42"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZCstguaruDzD"
      },
      "source": [
        "# Lectura de archivo\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MZ-i4EbmuIVX"
      },
      "source": [
        "### Desde google drive"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "9wJ_ERJKwzAQ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "bfcf65a3-951e-4714-9271-1187fadc5a96"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ]
        }
      ],
      "source": [
        "drive.mount('/content/drive')\n",
        "root_path = \"/content/drive/MyDrive/7506R-1C2024-GRUPO02/TP2/\""
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Aaf7SBZluLlI"
      },
      "source": [
        "### Desde maquina local"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "gPBHdD8KladB"
      },
      "outputs": [],
      "source": [
        "# Si se esta trabajando en local, descomentar esta celta y comentar la de arriba\n",
        "# root_path = \"./\""
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "DMJR5Rizlduf"
      },
      "outputs": [],
      "source": [
        "train_path = root_path + 'Dataset/train.csv'\n",
        "test_path = root_path + 'Dataset/test.csv'\n",
        "sample_path = root_path + 'Dataset/sample_solution.csv'\n",
        "ds_train = pd.read_csv(train_path)\n",
        "ds_test = pd.read_csv(test_path)\n",
        "sample = pd.read_csv(sample_path)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RlY6pAgdvbe5"
      },
      "source": [
        "# Visualizamos los datasets"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "cUWg64kmAV8w",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "1c5d5dae-9f87-4666-90c5-935d69555e12"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 50000 entries, 0 to 49999\n",
            "Data columns (total 3 columns):\n",
            " #   Column       Non-Null Count  Dtype \n",
            "---  ------       --------------  ----- \n",
            " 0   ID           50000 non-null  int64 \n",
            " 1   review_es    50000 non-null  object\n",
            " 2   sentimiento  50000 non-null  object\n",
            "dtypes: int64(1), object(2)\n",
            "memory usage: 1.1+ MB\n",
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 8599 entries, 0 to 8598\n",
            "Data columns (total 2 columns):\n",
            " #   Column     Non-Null Count  Dtype \n",
            "---  ------     --------------  ----- \n",
            " 0   ID         8599 non-null   int64 \n",
            " 1   review_es  8599 non-null   object\n",
            "dtypes: int64(1), object(1)\n",
            "memory usage: 134.5+ KB\n"
          ]
        }
      ],
      "source": [
        "ds_train.info()\n",
        "ds_test.info()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "KXiYdnPEvNU_",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 423
        },
        "outputId": "646dfe0a-08c7-4ebc-8f3f-d637d70a946a"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "          ID                                          review_es sentimiento\n",
              "0          0  Uno de los otros críticos ha mencionado que de...    positivo\n",
              "1          1  Una pequeña pequeña producción.La técnica de f...    positivo\n",
              "2          2  Pensé que esta era una manera maravillosa de p...    positivo\n",
              "3          3  Básicamente, hay una familia donde un niño peq...    negativo\n",
              "4          4  El \"amor en el tiempo\" de Petter Mattei es una...    positivo\n",
              "...      ...                                                ...         ...\n",
              "49995  49995  Pensé que esta película hizo un buen trabajo a...    positivo\n",
              "49996  49996  Mala parcela, mal diálogo, mala actuación, dir...    negativo\n",
              "49997  49997  Soy católica enseñada en escuelas primarias pa...    negativo\n",
              "49998  49998  Voy a tener que estar en desacuerdo con el com...    negativo\n",
              "49999  49999  Nadie espera que las películas de Star Trek se...    negativo\n",
              "\n",
              "[50000 rows x 3 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-e58f3337-31fe-4567-a4b0-24dae3be32b1\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>ID</th>\n",
              "      <th>review_es</th>\n",
              "      <th>sentimiento</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>Uno de los otros críticos ha mencionado que de...</td>\n",
              "      <td>positivo</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1</td>\n",
              "      <td>Una pequeña pequeña producción.La técnica de f...</td>\n",
              "      <td>positivo</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2</td>\n",
              "      <td>Pensé que esta era una manera maravillosa de p...</td>\n",
              "      <td>positivo</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>3</td>\n",
              "      <td>Básicamente, hay una familia donde un niño peq...</td>\n",
              "      <td>negativo</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>4</td>\n",
              "      <td>El \"amor en el tiempo\" de Petter Mattei es una...</td>\n",
              "      <td>positivo</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>49995</th>\n",
              "      <td>49995</td>\n",
              "      <td>Pensé que esta película hizo un buen trabajo a...</td>\n",
              "      <td>positivo</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>49996</th>\n",
              "      <td>49996</td>\n",
              "      <td>Mala parcela, mal diálogo, mala actuación, dir...</td>\n",
              "      <td>negativo</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>49997</th>\n",
              "      <td>49997</td>\n",
              "      <td>Soy católica enseñada en escuelas primarias pa...</td>\n",
              "      <td>negativo</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>49998</th>\n",
              "      <td>49998</td>\n",
              "      <td>Voy a tener que estar en desacuerdo con el com...</td>\n",
              "      <td>negativo</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>49999</th>\n",
              "      <td>49999</td>\n",
              "      <td>Nadie espera que las películas de Star Trek se...</td>\n",
              "      <td>negativo</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>50000 rows × 3 columns</p>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-e58f3337-31fe-4567-a4b0-24dae3be32b1')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-e58f3337-31fe-4567-a4b0-24dae3be32b1 button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-e58f3337-31fe-4567-a4b0-24dae3be32b1');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-268fe286-79bb-43ad-ae9e-fc448aca1849\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-268fe286-79bb-43ad-ae9e-fc448aca1849')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-268fe286-79bb-43ad-ae9e-fc448aca1849 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "\n",
              "  <div id=\"id_81995fcb-7da9-4ecb-a241-f0117376d5ba\">\n",
              "    <style>\n",
              "      .colab-df-generate {\n",
              "        background-color: #E8F0FE;\n",
              "        border: none;\n",
              "        border-radius: 50%;\n",
              "        cursor: pointer;\n",
              "        display: none;\n",
              "        fill: #1967D2;\n",
              "        height: 32px;\n",
              "        padding: 0 0 0 0;\n",
              "        width: 32px;\n",
              "      }\n",
              "\n",
              "      .colab-df-generate:hover {\n",
              "        background-color: #E2EBFA;\n",
              "        box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "        fill: #174EA6;\n",
              "      }\n",
              "\n",
              "      [theme=dark] .colab-df-generate {\n",
              "        background-color: #3B4455;\n",
              "        fill: #D2E3FC;\n",
              "      }\n",
              "\n",
              "      [theme=dark] .colab-df-generate:hover {\n",
              "        background-color: #434B5C;\n",
              "        box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "        filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "        fill: #FFFFFF;\n",
              "      }\n",
              "    </style>\n",
              "    <button class=\"colab-df-generate\" onclick=\"generateWithVariable('ds_train')\"\n",
              "            title=\"Generate code using this dataframe.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M7,19H8.4L18.45,9,17,7.55,7,17.6ZM5,21V16.75L18.45,3.32a2,2,0,0,1,2.83,0l1.4,1.43a1.91,1.91,0,0,1,.58,1.4,1.91,1.91,0,0,1-.58,1.4L9.25,21ZM18.45,9,17,7.55Zm-12,3A5.31,5.31,0,0,0,4.9,8.1,5.31,5.31,0,0,0,1,6.5,5.31,5.31,0,0,0,4.9,4.9,5.31,5.31,0,0,0,6.5,1,5.31,5.31,0,0,0,8.1,4.9,5.31,5.31,0,0,0,12,6.5,5.46,5.46,0,0,0,6.5,12Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "    <script>\n",
              "      (() => {\n",
              "      const buttonEl =\n",
              "        document.querySelector('#id_81995fcb-7da9-4ecb-a241-f0117376d5ba button.colab-df-generate');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      buttonEl.onclick = () => {\n",
              "        google.colab.notebook.generateWithVariable('ds_train');\n",
              "      }\n",
              "      })();\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "    </div>\n",
              "  </div>\n"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "dataframe",
              "variable_name": "ds_train",
              "summary": "{\n  \"name\": \"ds_train\",\n  \"rows\": 50000,\n  \"fields\": [\n    {\n      \"column\": \"ID\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 14433,\n        \"min\": 0,\n        \"max\": 49999,\n        \"num_unique_values\": 50000,\n        \"samples\": [\n          33553,\n          9427,\n          199\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"review_es\",\n      \"properties\": {\n        \"dtype\": \"string\",\n        \"num_unique_values\": 49599,\n        \"samples\": [\n          \"Dos a\\u00f1os despu\\u00e9s del \\u00e9xito de 'avi\\u00f3n', Jim Abrahams y Jerry y David Zucker crearon esta brillante comedia protagonizada por la Gran Leslie Nielsen, ya que el detective video simple 'Frank Drebin'. Tambi\\u00e9n en el elenco fue Alan North como 'Capit\\u00e1n Ed Hocken', Ed Williams como 'Ted Olsen' y William Duell como 'Johnny The Shoe Shine Boy'. '\\u00a1Escuadr\\u00f3n de la polic\\u00eda!' Destac\\u00f3 chistes sin verg\\u00fcenza y gags visuales inteligentes jugando en el fondo. Cada episodio concluir\\u00eda con un marco de congelaci\\u00f3n simulado en el que los caracteres en el marco se encuentran completamente quietos. \\u00a1Una de las mejores secuencias de 'congelamiento' de 'congelamiento' vieron a uno de los personajes que vierte caf\\u00e9 en una taza mientras est\\u00e1 de pie, lo que hace que la taza se desborde! Las estrellas invitadas fueron asesinadas en los t\\u00edtulos de apertura, \\u00a1uno incluido Georg Stanford Brown siendo aplastado por una caja fuerte! A pesar de obtener revisiones positivas y mucho cr\\u00edtico, '\\u00a1Escuadr\\u00f3n!' Solo dur\\u00f3 seis episodios antes de ser cancelados. Esto no signific\\u00f3 que el final, sin embargo, cinco a\\u00f1os despu\\u00e9s, el espect\\u00e1culo se transfiri\\u00f3 a la pantalla grande para la primera en la trilog\\u00eda de las pel\\u00edculas \\\"desnudas de la pistola\\\".\",\n          \"He le\\u00eddo los comentarios de KERCHING ON IMDB, y francamente, no he visto una revisi\\u00f3n positiva, hasta ahora.En realidad me gusta a Kerching.Kerching es sobre un adolescente llamado Taj Lewis, para hacer \\u00a3 1,000,000 para su madre, establece un sitio web llamado Rudeboy.El sitio web ofrece muchas cosas interesantes, aunque solo TAJ, y sus 2 amigos, Danny y Seymour pueden evitar que TAJ se exponga.Encuentro este espect\\u00e1culo bastante divertido y me gusta verlo.La actuaci\\u00f3n es O.K., pero definitivamente puede ser mejor, los personajes son divertidos, especialmente a Danny Spooner, el amigo de Taj, su estupidez es lo que lo hace divertido.y el otro amigo de Taj seymour.Tambi\\u00e9n Carlton, el due\\u00f1o de la cafeter\\u00eda llamada \\\"El fr\\u00edo\\\".Nunca lo vemos, solo lo escuchamos.El hermano menor de Taj, Omar, y su hermana mayor, Missy tambi\\u00e9n es divertida.Y la mejor amiga de Missy Kareesha.Muchas veces, Taj's casi se ha expuesto, casi.Adem\\u00e1s de la comedia, el drama en kerching, como la vida del amor y la p\\u00e9rdida de seres queridos hace que este programa sea excelente.\",\n          \"Una historia sobre el amor y el odio, la tragedia y la felicidad, y sobre todo, la amistad establecida en el momento muy interesante de la guerra civil estadounidense. Te interesa interesado en la historia, te involucra emocionalmente y te hace febrilmente esperar el pr\\u00f3ximo episodio., el casting era espl\\u00e9ndido.Muchas superestrellas aparecen en los camafeos cortos, los roles principales son jugados por una gran variedad de monumentos talentosos: Kirstie Alley y Terri Garber se deben mencionar aqu\\u00ed: este es simplemente un excelente ejemplo para una producci\\u00f3n de televisi\\u00f3n como deber\\u00eda ser. No se olvide de los pura.Lovelity of Wendy Kilbourne retratando constanza :-)\"\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"sentimiento\",\n      \"properties\": {\n        \"dtype\": \"category\",\n        \"num_unique_values\": 2,\n        \"samples\": [\n          \"negativo\",\n          \"positivo\"\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    }\n  ]\n}"
            }
          },
          "metadata": {},
          "execution_count": 38
        }
      ],
      "source": [
        "ds_train"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YrXa8bzno_t7"
      },
      "source": [
        "Dividimos el dataset de train en x_train e y_train (la primera tendra el texto y la segunda las labels).\n",
        "\n",
        "Adicionalmente no incluiremos la columna ID ya que no es necesaria para nuestro proposito de entrenamiento."
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def map_sentiment(ds, invert=False):\n",
        "  return ds.map({'negativo': 0, 'positivo': 1}) if not invert else ds.map({0: 'negativo', 1: 'positivo'})"
      ],
      "metadata": {
        "id": "OFsKvbWfQ-nq"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "T3Pz1qvoo_fM"
      },
      "outputs": [],
      "source": [
        "x_train_text_completo = ds_train[\"review_es\"]\n",
        "y_train_completo = map_sentiment(ds_train[\"sentimiento\"])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "1ZGZoNbcviQL",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 423
        },
        "outputId": "ea51423f-a499-47f0-9f2b-d2989259dd7d"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "         ID                                          review_es\n",
              "0     60000  La mayor virtud de esta película es su existen...\n",
              "1     60001  No soy un experto cinéfilo, pero pocas veces m...\n",
              "2     60002  Si no eres un incondicional del humor estilo T...\n",
              "3     60003  No sé qué está pasando, si la gente se deja ll...\n",
              "4     60004  Pero cuando amanece,y me quedo solo,siento en ...\n",
              "...     ...                                                ...\n",
              "8594  68594  Buena no, lo siguiente. Por fin un film serio ...\n",
              "8595  68595  Me esperaba mucho, pero que mucho, más.Guión m...\n",
              "8596  68596  De mal cuerpo como sensación al finalizar, de ...\n",
              "8597  68597  Los que han añadido comentarios os lo han dich...\n",
              "8598  68598  Fui a ver esta película de cine con entusiasmo...\n",
              "\n",
              "[8599 rows x 2 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-415b5dbc-0fe1-439b-b985-57e356de3937\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>ID</th>\n",
              "      <th>review_es</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>60000</td>\n",
              "      <td>La mayor virtud de esta película es su existen...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>60001</td>\n",
              "      <td>No soy un experto cinéfilo, pero pocas veces m...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>60002</td>\n",
              "      <td>Si no eres un incondicional del humor estilo T...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>60003</td>\n",
              "      <td>No sé qué está pasando, si la gente se deja ll...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>60004</td>\n",
              "      <td>Pero cuando amanece,y me quedo solo,siento en ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8594</th>\n",
              "      <td>68594</td>\n",
              "      <td>Buena no, lo siguiente. Por fin un film serio ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8595</th>\n",
              "      <td>68595</td>\n",
              "      <td>Me esperaba mucho, pero que mucho, más.Guión m...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8596</th>\n",
              "      <td>68596</td>\n",
              "      <td>De mal cuerpo como sensación al finalizar, de ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8597</th>\n",
              "      <td>68597</td>\n",
              "      <td>Los que han añadido comentarios os lo han dich...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8598</th>\n",
              "      <td>68598</td>\n",
              "      <td>Fui a ver esta película de cine con entusiasmo...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>8599 rows × 2 columns</p>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-415b5dbc-0fe1-439b-b985-57e356de3937')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-415b5dbc-0fe1-439b-b985-57e356de3937 button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-415b5dbc-0fe1-439b-b985-57e356de3937');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-c1b2a626-8057-4c9e-8fa4-2cc36d85b6ac\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-c1b2a626-8057-4c9e-8fa4-2cc36d85b6ac')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-c1b2a626-8057-4c9e-8fa4-2cc36d85b6ac button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "\n",
              "  <div id=\"id_e5d17808-67bd-498a-a569-1d11d1aa0f0a\">\n",
              "    <style>\n",
              "      .colab-df-generate {\n",
              "        background-color: #E8F0FE;\n",
              "        border: none;\n",
              "        border-radius: 50%;\n",
              "        cursor: pointer;\n",
              "        display: none;\n",
              "        fill: #1967D2;\n",
              "        height: 32px;\n",
              "        padding: 0 0 0 0;\n",
              "        width: 32px;\n",
              "      }\n",
              "\n",
              "      .colab-df-generate:hover {\n",
              "        background-color: #E2EBFA;\n",
              "        box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "        fill: #174EA6;\n",
              "      }\n",
              "\n",
              "      [theme=dark] .colab-df-generate {\n",
              "        background-color: #3B4455;\n",
              "        fill: #D2E3FC;\n",
              "      }\n",
              "\n",
              "      [theme=dark] .colab-df-generate:hover {\n",
              "        background-color: #434B5C;\n",
              "        box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "        filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "        fill: #FFFFFF;\n",
              "      }\n",
              "    </style>\n",
              "    <button class=\"colab-df-generate\" onclick=\"generateWithVariable('ds_test')\"\n",
              "            title=\"Generate code using this dataframe.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M7,19H8.4L18.45,9,17,7.55,7,17.6ZM5,21V16.75L18.45,3.32a2,2,0,0,1,2.83,0l1.4,1.43a1.91,1.91,0,0,1,.58,1.4,1.91,1.91,0,0,1-.58,1.4L9.25,21ZM18.45,9,17,7.55Zm-12,3A5.31,5.31,0,0,0,4.9,8.1,5.31,5.31,0,0,0,1,6.5,5.31,5.31,0,0,0,4.9,4.9,5.31,5.31,0,0,0,6.5,1,5.31,5.31,0,0,0,8.1,4.9,5.31,5.31,0,0,0,12,6.5,5.46,5.46,0,0,0,6.5,12Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "    <script>\n",
              "      (() => {\n",
              "      const buttonEl =\n",
              "        document.querySelector('#id_e5d17808-67bd-498a-a569-1d11d1aa0f0a button.colab-df-generate');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      buttonEl.onclick = () => {\n",
              "        google.colab.notebook.generateWithVariable('ds_test');\n",
              "      }\n",
              "      })();\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "    </div>\n",
              "  </div>\n"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "dataframe",
              "variable_name": "ds_test",
              "summary": "{\n  \"name\": \"ds_test\",\n  \"rows\": 8599,\n  \"fields\": [\n    {\n      \"column\": \"ID\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 2482,\n        \"min\": 60000,\n        \"max\": 68598,\n        \"num_unique_values\": 8599,\n        \"samples\": [\n          62499,\n          66723,\n          63944\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"review_es\",\n      \"properties\": {\n        \"dtype\": \"string\",\n        \"num_unique_values\": 8599,\n        \"samples\": [\n          \"Amenabar acert\\u00f3 de pleno con esta emotiva y dram\\u00e1tica historia. Como siempre en sus pel\\u00edculas, gran fotograf\\u00eda y banda sonora. Brillante Bardem y muy buenos secundarios. Totalmente recomendable, a no ser que s\\u00f3lo te guste el cine de acci\\u00f3n, si es as\\u00ed olv\\u00eddate.\",\n          \"Una de las obras espa\\u00f1olas m\\u00e1s exitosas del siglo XXI se trata en realidad de un coproducci\\u00f3n con Estados Unidos y M\\u00e9xico, de forma muy similar a Los Otros pero en esta ocasi\\u00f3n con actores nacionales. El resultado de esta producci\\u00f3n es el de una historia interesante, magn\\u00edficamente ambientada pero que flojea a partir de su mitad. El contraste entre la crueldad y la inhumanidad en la posguerra (aparcando debates pol\\u00edticos sobre su enfoque) y el mundo de fantas\\u00eda que s\\u00f3lo es capaz de ver la ni\\u00f1a resulta art\\u00edsticamente muy atractivo y diferente a lo que el cine espa\\u00f1ol nos tiene acostumbrados, pero el gui\\u00f3n resulta simple y pierde ritmo en su desarrollo.Sin embargo, las limitaciones argumentales son superadas por unas destacables interpretaciones de sus personajes. En especial de Sergi L\\u00f3pez, encarnando a un capit\\u00e1n despiadado, incapaz de sentir amor y obsesionado con eliminar a los que cree sus enemigos. Tambi\\u00e9n es digno de mencionar el trabajo siempre efectivo de Maribel Verd\\u00fa y las sorpresas de Doug Jones en el papel del fauno y la por aqu\\u00e9l entonces peque\\u00f1a Ivana Baquero, que ejecuta probablemente el mejor papel de su carrera hasta la fecha de la presente cr\\u00edtica.Sorprendentemente, la pel\\u00edcula se llev\\u00f3... \\u00a1tres Oscars! Y candidata para otros tres. Parece claro que para volver a triunfar en la gala de oro del cine hay que introducir recursos que admiren los americanos y adem\\u00e1s juntarnos con ellos.\",\n          \"Creo que se sobrevalora esta cinta, si bien es justo decir que la producci\\u00f3n es notable y un Luis Tosar no decepciona pero pienso que la trama empieza demasiado r\\u00e1pido no le da tiempo al espectador a hacerse con los personajes, los secundarios est\\u00e1n como fuera de lugar, alguno de ellos no veo que sean demasiado relevantes para la trama, est\\u00e1n como de paso, parece ser que los personajes principales arrastren el argumento,  creo que se ha desperdiciado un argumento muy original y se ha quedado en un intento bastante flojo.\"\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    }\n  ]\n}"
            }
          },
          "metadata": {},
          "execution_count": 41
        }
      ],
      "source": [
        "ds_test"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IGXF2ESLqFR2"
      },
      "source": [
        "Creamos el dataset de x_test sin incluir el id"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "EWVFwP9JqK-6"
      },
      "outputs": [],
      "source": [
        "x_test_text = ds_test[\"review_es\"]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cCvNYz6zpT9S"
      },
      "source": [
        "# Preprocesamiento"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "J3IUR3SjQ_ep"
      },
      "source": [
        "## Division de train y validation\n",
        "\n",
        "Dividimos el training set en training y validating sets"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "1EUYD4uWo0sU"
      },
      "outputs": [],
      "source": [
        "x_train_text, x_validation_text, y_train, y_validation = train_test_split(\n",
        "    ds_train['review_es'], ds_train[\"sentimiento\"],\n",
        "    stratify=ds_train[\"sentimiento\"],\n",
        "    test_size=0.20,\n",
        "    random_state=RAND_SEED\n",
        "    )"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "y_train = map_sentiment(y_train)"
      ],
      "metadata": {
        "id": "Ko82SPl_a1Kn"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "p3QxP7yhoWcS"
      },
      "source": [
        "## Normalizacion"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FFvyjNm7QOZR"
      },
      "source": [
        "### Tokenizacion"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ldHkUQ_JoVs-"
      },
      "outputs": [],
      "source": [
        "def clean_text(text):\n",
        "  return text.strip()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "H77tYj-hofDr",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "efcd4867-9299-474d-e290-75fb9efab8cd"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['No',\n",
              " 'puedo',\n",
              " 'creer',\n",
              " 'que',\n",
              " 'me',\n",
              " 'permita',\n",
              " 'a',\n",
              " 'esta',\n",
              " 'película',\n",
              " 'para']"
            ]
          },
          "metadata": {},
          "execution_count": 46
        }
      ],
      "source": [
        "def tokenizer(text):\n",
        "  return [w for w in word_tokenize(text) if w.isalpha()]\n",
        "\n",
        "tokenizer(clean_text(x_train_text.iloc[1]))[:10]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5Fi074MHohjS"
      },
      "source": [
        "Defino un stop list de palabras en español"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "UzeKYZpvolnz",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5c43e224-0117-4d88-e453-5f0626784993"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['de',\n",
              " 'la',\n",
              " 'que',\n",
              " 'el',\n",
              " 'en',\n",
              " 'y',\n",
              " 'a',\n",
              " 'los',\n",
              " 'del',\n",
              " 'se',\n",
              " 'las',\n",
              " 'por',\n",
              " 'un',\n",
              " 'para',\n",
              " 'con']"
            ]
          },
          "metadata": {},
          "execution_count": 47
        }
      ],
      "source": [
        "stoplist = stopwords.words(\"spanish\")\n",
        "stoplist[:15]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aBecHNdbreQF"
      },
      "source": [
        "Tokenizamos las stopwords con el mismo tokenizador que utilizamos en el corpus"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Ai9c1A83rZCs"
      },
      "outputs": [],
      "source": [
        "stoplist_tokenized = []\n",
        "\n",
        "for w in stoplist:\n",
        "    stoplist_tokenized = stoplist_tokenized + tokenizer(w)\n",
        "\n",
        "stoplist_tokenized = list(set(stoplist_tokenized)) # sin repedidos"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DHwbnsuBQQ8X"
      },
      "source": [
        "### Stem - Lematizacion"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "slorXOv3vwNP"
      },
      "source": [
        "Visualizamos qué tipo de palabras tenemos en train"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "FkIZ5cSnv2gG",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "7ca99b89-e7e7-4074-95f8-78caeaaf8b58"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[('No', 'DT'),\n",
              " ('puedo', 'NN'),\n",
              " ('creer', 'NN'),\n",
              " ('que', 'IN'),\n",
              " ('me', 'PRP'),\n",
              " ('permita', 'VBP'),\n",
              " ('a', 'DT'),\n",
              " ('esta', 'JJ'),\n",
              " ('película', 'NN'),\n",
              " ('para', 'NN')]"
            ]
          },
          "metadata": {},
          "execution_count": 49
        }
      ],
      "source": [
        "postags = pos_tag(word_tokenize(clean_text(x_train_text.iloc[1])))\n",
        "postags[:10]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7hlLs7Hgv7RW"
      },
      "source": [
        "Podemos ver que nos encontramos con palabras como adverbios, verbos, preposiciones, adjetivos, pronombres, sustantivos, etc."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Pw0G01XPAb-J"
      },
      "source": [
        "Instanciamos un lemmatizador"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "rJTkhcyyAbP-"
      },
      "outputs": [],
      "source": [
        "wnl = WordNetLemmatizer()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GuWpA1-2wAhc"
      },
      "source": [
        "Generamos la función de lematización por pos\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Ce2P-TyEv_64"
      },
      "outputs": [],
      "source": [
        "def word_pos_to_lemma(word, pos, wnl):\n",
        "    if pos.startswith('J'):\n",
        "        return wnl.lemmatize(word,wordnet.ADJ) # adjetivo\n",
        "    elif pos.startswith('V'):\n",
        "        return wnl.lemmatize(word,wordnet.VERB) # vervo\n",
        "    elif pos.startswith('N'):\n",
        "        return wnl.lemmatize(word,wordnet.NOUN)# sustantivo\n",
        "    elif pos.startswith('R'):\n",
        "        return wnl.lemmatize(word,wordnet.ADV) # advervio\n",
        "    else:\n",
        "        return wnl.lemmatize(word)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "YAWLvjHmwDbV",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "89099450-c05a-4647-a024-a33dadd1f4f5"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[('No', 'No'),\n",
              " ('puedo', 'puedo'),\n",
              " ('creer', 'creer'),\n",
              " ('que', 'que'),\n",
              " ('me', 'me'),\n",
              " ('permita', 'permita'),\n",
              " ('a', 'a'),\n",
              " ('esta', 'esta'),\n",
              " ('película', 'película'),\n",
              " ('para', 'para')]"
            ]
          },
          "metadata": {},
          "execution_count": 52
        }
      ],
      "source": [
        "lemmatized_words = [(w, word_pos_to_lemma(w, pos, wnl)) for w, pos in postags if w.isalpha()]\n",
        "lemmatized_words[:10]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "B0cCIAacwG9T",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "63a309b4-938b-4bec-ca61-6468998b5523"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[(('No', 'No'), 1),\n",
              " (('puedo', 'puedo'), 1),\n",
              " (('creer', 'creer'), 1),\n",
              " (('que', 'que'), 4),\n",
              " (('me', 'me'), 3),\n",
              " (('permita', 'permita'), 1),\n",
              " (('a', 'a'), 8),\n",
              " (('esta', 'esta'), 2),\n",
              " (('película', 'película'), 3),\n",
              " (('para', 'para'), 4)]"
            ]
          },
          "metadata": {},
          "execution_count": 53
        }
      ],
      "source": [
        "# Usar Counter para contar las repeticiones de cada tupla\n",
        "lemmatized_counter = Counter(lemmatized_words)\n",
        "\n",
        "# Imprimir los resultados\n",
        "list(lemmatized_counter.items())[:10]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5NkN1dWuwNLV"
      },
      "source": [
        "Podemos ver que las lematizaciones se aplican más de una vez"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "epzc22BowLaM"
      },
      "outputs": [],
      "source": [
        "def lemmatize_text(text):\n",
        "    lemmatizer = WordNetLemmatizer()\n",
        "    words = word_tokenize(text)\n",
        "    lemmatized_words = [lemmatizer.lemmatize(word) for word in words if word.isalpha()]\n",
        "    return lemmatized_words"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "DojmiyyzwQGu",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a163fbdc-7b89-4315-d3dd-b8193e0534b7"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['No',\n",
              " 'puedo',\n",
              " 'creer',\n",
              " 'que',\n",
              " 'me',\n",
              " 'permita',\n",
              " 'a',\n",
              " 'esta',\n",
              " 'película',\n",
              " 'para']"
            ]
          },
          "metadata": {},
          "execution_count": 55
        }
      ],
      "source": [
        "lemmatize_text(clean_text(x_train_text.iloc[1]))[:10]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zd9swgrXpl5r"
      },
      "source": [
        "## Vectorizacion"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mQSwdSfeprwe"
      },
      "source": [
        "### Bag of words"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QJdXIdhwp1YF"
      },
      "source": [
        "Con tokenizacion sin lemmas"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "p067t0_8p0HH",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d8dab571-69d2-470d-dc91-f740e01fd093"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<40000x48415 sparse matrix of type '<class 'numpy.int64'>'\n",
              "\twith 3999252 stored elements in Compressed Sparse Row format>"
            ]
          },
          "metadata": {},
          "execution_count": 59
        }
      ],
      "source": [
        "count_vect = CountVectorizer(preprocessor=clean_text,\n",
        "                             tokenizer=tokenizer,\n",
        "                             min_df=5,\n",
        "                             stop_words=stoplist_tokenized)\n",
        "x_train_bow_tok = count_vect.fit_transform(x_train_text) # cuenta frecuencia de tokens y define el diccionario\n",
        "x_validation_bow_tok = count_vect.transform(x_validation_text) # cuenta frecuencia de tokens existentes en el diccionario\n",
        "x_train_bow_tok"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WmUxNdi5p9kS"
      },
      "source": [
        "Con tokenizacion con lemmas"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "64Uiu6iQqGhv",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "0adfb796-8222-48b6-f9c2-10f35c52d051"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<40000x47766 sparse matrix of type '<class 'numpy.int64'>'\n",
              "\twith 5308134 stored elements in Compressed Sparse Row format>"
            ]
          },
          "metadata": {},
          "execution_count": 60
        }
      ],
      "source": [
        "count_vect = CountVectorizer(preprocessor=clean_text,\n",
        "                             tokenizer=lemmatize_text,\n",
        "                             min_df=5,\n",
        "                             stop_words=lemmatized_words)\n",
        "x_train_bow_lemma = count_vect.fit_transform(x_train_text) # cuenta frecuencia de tokens y define el diccionario\n",
        "x_validation_bow_lemma = count_vect.transform(x_validation_text) # cuenta frecuencia de tokens existentes en el diccionario\n",
        "x_train_bow_lemma"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hGhVmwlPp-EO"
      },
      "source": [
        "### N-gramas"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3JpyD5VLqMp-"
      },
      "source": [
        "Con tokenizacion sin lemmas"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "I_VRagvvqNLR",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a6fa7e8a-2834-4a5c-d6ca-c8375605229b"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<40000x152699 sparse matrix of type '<class 'numpy.int64'>'\n",
              "\twith 5610823 stored elements in Compressed Sparse Row format>"
            ]
          },
          "metadata": {},
          "execution_count": 61
        }
      ],
      "source": [
        "count_vect = CountVectorizer(preprocessor=clean_text,\n",
        "                             tokenizer=tokenizer,\n",
        "                             min_df=5,\n",
        "                             stop_words=stoplist_tokenized,\n",
        "                             ngram_range=(1,2))\n",
        "x_train_ngram_tok = count_vect.fit_transform(x_train_text) # cuenta frecuencia de tokens y define el diccionario\n",
        "x_validation_ngram_tok = count_vect.transform(x_validation_text) # cuenta frecuencia de tokens existentes en el diccionario\n",
        "x_train_ngram_tok"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ff9kzqcgqUCT"
      },
      "source": [
        "Con tokenizacion con lemmas"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "-lCLo98jqUZN",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "dc9989df-fe0f-4638-bfb0-8de43cd7683d"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<40000x231974 sparse matrix of type '<class 'numpy.int64'>'\n",
              "\twith 11406128 stored elements in Compressed Sparse Row format>"
            ]
          },
          "metadata": {},
          "execution_count": 62
        }
      ],
      "source": [
        "count_vect = CountVectorizer(preprocessor=clean_text,\n",
        "                             tokenizer=lemmatize_text,\n",
        "                             min_df=5,\n",
        "                             stop_words=lemmatized_words,\n",
        "                             ngram_range=(1,2))\n",
        "x_train_ngram_lemma = count_vect.fit_transform(x_train_text) # cuenta frecuencia de tokens y define el diccionario\n",
        "x_validation_ngram_lemma = count_vect.transform(x_validation_text) # cuenta frecuencia de tokens existentes en el diccionario\n",
        "x_train_ngram_lemma"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "k0pO62VZp_ig"
      },
      "source": [
        "### TF-IDF"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gYe-ZNjBqbGj"
      },
      "source": [
        "Con tokenizacion sin lemmas"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "y42CUXshqbd5",
        "outputId": "d5862356-aa38-4567-a46b-d3660e89eec8"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<40000x48415 sparse matrix of type '<class 'numpy.float64'>'\n",
              "\twith 3999252 stored elements in Compressed Sparse Row format>"
            ]
          },
          "metadata": {},
          "execution_count": 63
        }
      ],
      "source": [
        "tfidf_vect = TfidfVectorizer(\n",
        "    preprocessor=clean_text,\n",
        "    tokenizer=tokenizer,\n",
        "    min_df=5,\n",
        "    stop_words=stoplist_tokenized\n",
        ")\n",
        "x_train_tfidf_tok = tfidf_vect.fit_transform(x_train_text) # cuenta frecuencia de tokens y define el diccionario\n",
        "x_validation_tfidf_tok = tfidf_vect.transform(x_validation_text) # cuenta frecuencia de tokens existentes en el diccionario\n",
        "x_train_tfidf_tok"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8Au-_UudqkDa"
      },
      "source": [
        "Con tokenizacion con lemmas"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "COH2xIjXqmKw",
        "outputId": "b622175d-88c3-4b7e-97aa-4b180dc0a33d"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<40000x47766 sparse matrix of type '<class 'numpy.float64'>'\n",
              "\twith 5308134 stored elements in Compressed Sparse Row format>"
            ]
          },
          "metadata": {},
          "execution_count": 64
        }
      ],
      "source": [
        "tfidf_vect = TfidfVectorizer(\n",
        "    preprocessor=clean_text,\n",
        "    tokenizer=lemmatize_text,\n",
        "    min_df=5,\n",
        "    stop_words=lemmatized_words\n",
        ")\n",
        "x_train_tfidf_lemma = tfidf_vect.fit_transform(x_train_text)\n",
        "x_validation_tfidf_lemma = tfidf_vect.transform(x_validation_text)\n",
        "x_train_tfidf_lemma"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PdlbIp4EBQVM"
      },
      "source": [
        "### Utils\n",
        "\n",
        "Creacion de algunas funciones de utilidad"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "1w58Vv8VBWlt"
      },
      "outputs": [],
      "source": [
        "def plot_confussion_matrix(true, pred):\n",
        "  #Creo matriz de confusión\n",
        "  tabla = confusion_matrix(true, pred)\n",
        "\n",
        "  #Grafico matriz de confusión\n",
        "  sns.heatmap(tabla, cmap='Blues',annot=True,fmt='g')\n",
        "  plt.xlabel('Predicciones')\n",
        "  plt.ylabel('Verdaderos')\n",
        "\n",
        "  #Reporte\n",
        "  print(classification_report(true, pred))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "IJd51iWFDGZ1"
      },
      "outputs": [],
      "source": [
        "def build_results(pred):\n",
        "  res = ds_test.copy()\n",
        "  res.drop(columns=[\"review_es\"], inplace=True)\n",
        "  res['sentimiento'] = pred\n",
        "  return res"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "tciD9NNPDm-C"
      },
      "outputs": [],
      "source": [
        "def print_positive_negatives(sentimientos):\n",
        "  total_positives = (sentimientos == \"positivo\").sum()\n",
        "  total_negatives = (sentimientos == \"negativo\").sum()\n",
        "  print(f\"Positives: {total_positives}\\nNegatives: {total_negatives}\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "20fI2ToCMG_J"
      },
      "source": [
        "# Naive Bayes"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bMK-EtU5rtKt"
      },
      "source": [
        "### Modelos\n",
        "\n",
        "Vemos como performa el mismo modelo con diferentes normalizaciones y vectorizacion"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jrG2Tl1xWhGK"
      },
      "source": [
        "Bag of words Con tokenizacion sin lemmas"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 615
        },
        "id": "54N_D4FDWqEz",
        "outputId": "c6e22f40-4342-4b0c-9307-b568173cbbfb"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "    negativo       0.82      0.87      0.85      5000\n",
            "    positivo       0.86      0.81      0.84      5000\n",
            "\n",
            "    accuracy                           0.84     10000\n",
            "   macro avg       0.84      0.84      0.84     10000\n",
            "weighted avg       0.84      0.84      0.84     10000\n",
            "\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "clf = MultinomialNB(alpha=1).fit(x_train_bow_tok, y_train) # entreno\n",
        "y_pred = clf.predict(x_validation_bow_tok) # predigo\n",
        "y_pred = map_sentiment(pd.Series(y_pred), invert=True)\n",
        "plot_confussion_matrix(y_validation, y_pred) # grafico"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rtH9RxphWiqX"
      },
      "source": [
        "Bag of words Con tokenizacion con lemmas"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 615
        },
        "id": "sBZjXZdrWxXC",
        "outputId": "2c57e528-2337-4ecb-becc-8d865668728b"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "    negativo       0.81      0.87      0.84      5000\n",
            "    positivo       0.86      0.80      0.83      5000\n",
            "\n",
            "    accuracy                           0.83     10000\n",
            "   macro avg       0.84      0.83      0.83     10000\n",
            "weighted avg       0.84      0.83      0.83     10000\n",
            "\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "clf = MultinomialNB(alpha=1).fit(x_train_bow_lemma, y_train)\n",
        "y_pred = clf.predict(x_validation_bow_lemma)\n",
        "y_pred = map_sentiment(pd.Series(y_pred), invert=True)\n",
        "plot_confussion_matrix(y_validation, y_pred)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_IX0Y61FV011"
      },
      "source": [
        "N-gramas Con tokenizacion sin lemmas"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 615
        },
        "id": "KdwjjYxJXAmr",
        "outputId": "3b2b72f1-cc89-482b-cc0f-daee3ae712a4"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "    negativo       0.86      0.88      0.87      5000\n",
            "    positivo       0.88      0.86      0.87      5000\n",
            "\n",
            "    accuracy                           0.87     10000\n",
            "   macro avg       0.87      0.87      0.87     10000\n",
            "weighted avg       0.87      0.87      0.87     10000\n",
            "\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "clf = MultinomialNB(alpha=1).fit(x_train_ngram_tok, y_train)\n",
        "y_pred = clf.predict(x_validation_ngram_tok)\n",
        "y_pred = map_sentiment(pd.Series(y_pred), invert=True)\n",
        "plot_confussion_matrix(y_validation, y_pred)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JN_cRdpyV3l8"
      },
      "source": [
        "N-gramas Con tokenizacion con lemmas"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 615
        },
        "id": "UoEWyLSpXKkH",
        "outputId": "df21419c-4753-44bf-c812-79ee7df00980"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "    negativo       0.85      0.87      0.86      5000\n",
            "    positivo       0.87      0.85      0.86      5000\n",
            "\n",
            "    accuracy                           0.86     10000\n",
            "   macro avg       0.86      0.86      0.86     10000\n",
            "weighted avg       0.86      0.86      0.86     10000\n",
            "\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "clf = MultinomialNB(alpha=1).fit(x_train_ngram_lemma, y_train)\n",
        "y_pred = clf.predict(x_validation_ngram_lemma)\n",
        "y_pred = map_sentiment(pd.Series(y_pred), invert=True)\n",
        "plot_confussion_matrix(y_validation, y_pred)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pkltW2KDa8GM"
      },
      "source": [
        "Tf-idf Con tokenizacion sin lemmas"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 615
        },
        "id": "MLHS_7_aa0PR",
        "outputId": "1f3c27ee-1455-4409-e4d8-74f90b2cf5d4"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "    negativo       0.85      0.86      0.85      5000\n",
            "    positivo       0.86      0.85      0.85      5000\n",
            "\n",
            "    accuracy                           0.85     10000\n",
            "   macro avg       0.85      0.85      0.85     10000\n",
            "weighted avg       0.85      0.85      0.85     10000\n",
            "\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "clf = MultinomialNB(alpha=1).fit(x_train_tfidf_tok, y_train)\n",
        "y_pred = clf.predict(x_validation_tfidf_tok)\n",
        "y_pred = map_sentiment(pd.Series(y_pred), invert=True)\n",
        "plot_confussion_matrix(y_validation, y_pred)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "g1N8xz3muwy3"
      },
      "source": [
        "Tf-idf Con tokenizacion con lemmas"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 615
        },
        "id": "1-SJ9qiwu03g",
        "outputId": "2e410d78-0cc8-475e-f100-ee1a1acf7c9c"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "    negativo       0.85      0.86      0.85      5000\n",
            "    positivo       0.86      0.84      0.85      5000\n",
            "\n",
            "    accuracy                           0.85     10000\n",
            "   macro avg       0.85      0.85      0.85     10000\n",
            "weighted avg       0.85      0.85      0.85     10000\n",
            "\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 2 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiQAAAGwCAYAAACZ7H64AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAABCgklEQVR4nO3df3zN9f//8fsZ2zFsY2O/3n4kKyxGplhFZJma4t364ZNQRDTEhFZSlOZNEhG9U029KSq/FQ1twvxo3svvhearYvMr1hb7Yef7h7fz7kSdHe/z8pq5Xbu8Lhfn+Xqe53m8VvTweDxfr2Ox2Ww2AQAAmMjD7AAAAABISAAAgOlISAAAgOlISAAAgOlISAAAgOlISAAAgOlISAAAgOlISAAAgOkqmx2AEbxvHmR2CEC59MvW6WaHAJQ7Va7A/wnd9f+lM/+uuL+HqZAAAADTVcgKCQAA5YqFv/87Q0ICAIDRLBazIyj3SEgAADAaFRKn+AkBAADTUSEBAMBotGycIiEBAMBotGyc4icEAABMR4UEAACj0bJxioQEAACj0bJxip8QAAAwHRUSAACMRsvGKRISAACMRsvGKX5CAADAdFRIAAAwGi0bp0hIAAAwGi0bp0hIAAAwGhUSp0jZAACA6aiQAABgNFo2TpGQAABgNBISp/gJAQAA01EhAQDAaB5sanWGhAQAAKPRsnGKnxAAADAdFRIAAIzGc0icIiEBAMBotGyc4icEAMA1YMKECbJYLBo6dKh97OzZs4qPj1dAQICqV6+uuLg45ebmOrzv0KFDio2NVdWqVRUYGKgRI0aopKTEYU5qaqpatmwpq9WqsLAwJScnuxwfCQkAAEazWNxzXKatW7fqnXfeUUREhMP4sGHDtGzZMn366adKS0vT4cOH9cADD9jPnzt3TrGxsSoqKtLGjRs1Z84cJScna8yYMfY52dnZio2NVYcOHZSZmamhQ4fqySef1KpVq1yKkYQEAACjWTzcc1yG/Px89ejRQ++++65q1qxpHz99+rTee+89vfHGG7rrrrsUGRmpDz74QBs3btSmTZskSV999ZV2796tf/3rX2rRooXuuecevfLKK5oxY4aKiookSbNmzVKDBg00efJkNWnSRIMGDdKDDz6oKVOmuBQnCQkAAEZzU4WksLBQeXl5DkdhYeFffnR8fLxiY2MVHR3tMJ6RkaHi4mKH8caNG6tevXpKT0+XJKWnp6tZs2YKCgqyz4mJiVFeXp527dpln/PHtWNiYuxrlBUJCQAAV4mkpCT5+fk5HElJSX86/5NPPtG2bdsuOScnJ0deXl6qUaOGw3hQUJBycnLsc36fjFw4f+HcX83Jy8vTmTNnynxt3GUDAIDR3HSXTWJiohISEhzGrFbrJef++OOPeuaZZ5SSkqIqVaq45fONRIUEAACjuallY7Va5evr63D8WUKSkZGho0ePqmXLlqpcubIqV66stLQ0TZs2TZUrV1ZQUJCKiop06tQph/fl5uYqODhYkhQcHHzRXTcXXjub4+vrK29v7zL/iEhIAACogDp27KgdO3YoMzPTfrRq1Uo9evSw/9rT01Nr1qyxvycrK0uHDh1SVFSUJCkqKko7duzQ0aNH7XNSUlLk6+ur8PBw+5zfr3FhzoU1yoqWDQAARjPhwWg+Pj5q2rSpw1i1atUUEBBgH+/bt68SEhLk7+8vX19fDR48WFFRUWrTpo0kqVOnTgoPD1fPnj01ceJE5eTkaPTo0YqPj7dXZgYMGKDp06dr5MiR6tOnj9auXasFCxZoxYoVLsVLQgIAgNHK6aPjp0yZIg8PD8XFxamwsFAxMTF6++237ecrVaqk5cuXa+DAgYqKilK1atXUu3dvjRs3zj6nQYMGWrFihYYNG6apU6eqTp06mj17tmJiYlyKxWKz2Wxuu7JywvvmQWaHAJRLv2ydbnYIQLlT5Qr81dw7dppb1jmzYohb1imPqJAAAGA0vsvGKRISAACMRkLiFD8hAABgOiokAAAYrZxuai1PSEgAADAaLRunSEgAADAaFRKnSNkAAIDpqJAAAGA0WjZOkZAAAGA0WjZOkbIBAADTUSEBAMBgFiokTpGQAABgMBIS52jZAAAA01EhAQDAaBRInCIhAQDAYLRsnKNlAwAATEeFBAAAg1EhcY6EBAAAg5GQOEdCAgCAwUhInGMPCQAAMB0VEgAAjEaBxCkSEgAADEbLxjlaNgAAwHRUSAAAMBgVEudISAAAMBgJiXO0bAAAgOmokAAAYDAqJM6RkAAAYDTyEado2QAAANNRIQEAwGC0bJwjIQEAwGAkJM6RkAAAYDASEufYQwIAAExHhQQAAKNRIHGKhAQAAIPRsnGOlg0AADAdFRIAAAxGhcQ5EhIAAAxGQuIcLRsAAGA6KiQAABiMColzJCQAABiNfMQpWjYAAFRAM2fOVEREhHx9feXr66uoqCh9+eWX9vPt27eXxWJxOAYMGOCwxqFDhxQbG6uqVasqMDBQI0aMUElJicOc1NRUtWzZUlarVWFhYUpOTr6seKmQAABgMDNaNnXq1NGECRN0ww03yGazac6cOeratav+/e9/66abbpIk9evXT+PGjbO/p2rVqvZfnzt3TrGxsQoODtbGjRt15MgR9erVS56ennrttdckSdnZ2YqNjdWAAQM0d+5crVmzRk8++aRCQkIUExPjUrwWm81mc8N1lyveNw8yOwSgXPpl63SzQwDKnSpX4K/mdZ5e7JZ1Dky5R4WFhQ5jVqtVVqu1TO/39/fXpEmT1LdvX7Vv314tWrTQm2++ecm5X375pbp06aLDhw8rKChIkjRr1iyNGjVKx44dk5eXl0aNGqUVK1Zo586d9vd1795dp06d0sqVK126Nlo2AAAY7I+tkcs9kpKS5Ofn53AkJSU5/fxz587pk08+UUFBgaKiouzjc+fOVa1atdS0aVMlJibqt99+s59LT09Xs2bN7MmIJMXExCgvL0+7du2yz4mOjnb4rJiYGKWnp7v8M6JlAwDAVSIxMVEJCQkOY39VHdmxY4eioqJ09uxZVa9eXYsWLVJ4eLgk6dFHH1X9+vUVGhqq7du3a9SoUcrKytLChQslSTk5OQ7JiCT765ycnL+ck5eXpzNnzsjb27vM10ZCAgCA0dy0hcSV9owkNWrUSJmZmTp9+rQ+++wz9e7dW2lpaQoPD1f//v3t85o1a6aQkBB17NhRBw4cUMOGDd0TsAto2QAAYDB3tWxc5eXlpbCwMEVGRiopKUnNmzfX1KlTLzm3devWkqT9+/dLkoKDg5Wbm+sw58Lr4ODgv5zj6+vrUnVEIiEBAOCaUVpaetGm2AsyMzMlSSEhIZKkqKgo7dixQ0ePHrXPSUlJka+vr73tExUVpTVr1jisk5KS4rBPpaxISOCSZ5+4W2f+PV2Tno2zj731QnftWvqSTqa/oUNrk7RgSn/deJ1jT7FucE0tnDZAJza+of+3JkmvDe2mSpUc//NrG3mDNs4bpVObp2jnkpf02H2tr8g1Ae5wz913qflNjS46XntlrCSpsLBQr70yVu1ua602rW5WwjODdeL4cYc1jhw+rEED+6t1ZHO1bxulN17/x0XPfMDVyYwKSWJiotatW6eDBw9qx44dSkxMVGpqqnr06KEDBw7olVdeUUZGhg4ePKilS5eqV69eateunSIiIiRJnTp1Unh4uHr27KnvvvtOq1at0ujRoxUfH29vGw0YMEA//PCDRo4cqb179+rtt9/WggULNGzYMJd/RuwhQZlFhtdT37jbtf37nxzG/73nR33y5Vb9eOQX+ftV1QsDYrX87Xg17vKSSktt8vCwaOG0gco9kacOj09WcG0/zX6lp4pLzuml6cskSfVDA7TorQGa/dl6PfFCsjrc2kgzxzyqnON5Wp2+x4zLBVwyd/5nKj13zv56//59eurJJ3R3TGdJ0qR/vKZv0tI06Y035ePjo6TxryjhmUGaM/cTSefvghj09FOqVauW5vzrEx0/flSjE0epcmVPDRmacMnPxNXDjOeQHD16VL169dKRI0fk5+eniIgIrVq1Snfffbd+/PFHrV69Wm+++aYKCgpUt25dxcXFafTo0fb3V6pUScuXL9fAgQMVFRWlatWqqXfv3g7PLWnQoIFWrFihYcOGaerUqapTp45mz57t8jNIJJ5DgjKq5u2l9I+f0zNJ8/Xck521PesnjXj980vObXpDqLYueF7h972s7J+Oq9Pt4Vo4dYCu7/SCjp78VZL05IN36NUhXVX3rudUXHJOrw7pqs5tb1Krh16zr/PhhCfkV91bXQe9fUWu8VrAc0iunIlJ47UuLVXLvvxK+fn5an9HlCZMfN2eoGT/cEDd7rtXH82br4jmLbT+mzQNfnqAVn/9jQJq1ZIkLZj/saa+8bpSv0mXp5eXmZdToV2J55Bc98xyt6xzcGoXt6xTHtGyQZm8mfiIVn6zU19vzvrLeVWreKnX/W2U/dNx/ZTziySpdUQD7dx/2J6MSFLKxj3y8/FWeMPzvcrWzRtctHbKxj1qHdHAzVcCGK+4qEgrli9VtwfiZLFYtHvXTpWUFKt11G32OQ2ub6iQkFB995++/XeZmbrhhhvtyYgk3Xb7HcrPz9f+A/uv9CXAzcza1Ho1MbVlc/z4cb3//vtKT0+339McHBys2267TY8//rhq165tZnj4j4diItWicV3d8djEP53T/6G2Gj+0m6pXtSorO0exA6eruOR8+ToowFdHT/zqMP/oybzz52r5Slnn5+SevHiOn4+3qlg9dbaw2M1XBRhn7drV+vXXX3V/t79Lkk4cPy5PT0/5+vo6zPMPCNDx48fsc/wDajmcD/jP6xP/mYOrWMXOJdzCtArJ1q1bdeONN2ratGny8/NTu3bt1K5dO/n5+WnatGlq3Lixvv32W6frFBYWKi8vz+GwlZ5z+j6UTZ2gGpo0Ik5PvJCswqI/31z3yZdb1eb/Jii67xTtO3RM//pHH1m92KKEa9Oizz/X7Xe0U2BgkPPJACSZWCEZPHiwHnroIc2aNeuiMpTNZtOAAQM0ePBgp4+fTUpK0tixYx3GKgXdIs+QW90e87Xo5ib1FBTgq/R5o+xjlStX0h0tG2rAI+3k13qoSkttyss/q7z8szpw6Ji2bD+oI+smqutdzbVgZYZyT+SpVdP6DusG+p//m2Lu8fOVktwTeQry97lozulfz1AdwVXl8OGftXnTRr0x9S37WECtWiouLlZeXp5DleTkiROqVau2fc7OHdsd1jpx4vh/zlEtvtpV9HaLO5hWIfnuu+80bNiwS/5LslgsGjZsmP2e6L+SmJio06dPOxyVgyINiPja9PWWLEU+OF6tu0+wHxm7/p8++eJbte4+QaWlF++JtlgsssgiL8/z+e7m7dlqGhaq2jWr2+d0bNNYp389oz0/nG/Vbf4uW+1vbeSwTsc2jbV5e7aBVwe435JFC+XvH6C27drbx8JvaqrKlT21ZdN//4J1MPsHHTlyWM1btJAkNW/RQvv2fa8TJ07Y52zauFHVq1dXw4ZhVyp8GIQ9JM6ZViEJDg7Wli1b1Lhx40ue37Jly0XPx7+USz1G1+JRyS0xQsr/rVC7DxxxGCs4U6STpwu0+8ARXfe3AD0YE6k16Xt0/Jd8/S2ohoY/0UlnCou1av35L19anb5He37I0Xuv9tYLUxcrKMBXL8V30TsL1qmo+Hwb6N3P1mtA93Ya/0xXzVmySe1vuVFxd9+svw+ZdcWvGbhcpaWlWrJooe7r2k2VK//3j1cfHx/9PS5Or0+cIF8/P1WvXl0TXntVzVvcrIjmLSRJUbfdoesbhumF50Zq2PAROn78mKa/9aYe+b8e8uIOm6teBc8l3MK0hOTZZ59V//79lZGRoY4dO9qTj9zcXK1Zs0bvvvuuXn/9dbPCQxkVFpXo9psbatCj7VXTt6qOnvhV67ftV4fHJ+vYL/mSpNJSm+Kemampz3dXavJwFZwt1NxlWzRu5gr7Ov/v8An9ffAsTXz2AcU/2l4/557SwHHzeAYJriqb0jfqyJHD6vZA3EXnRox6Xh4WDw0fOkRFxUW67fY79MLol+znK1WqpLfenqXx415Wrx6PyNvbW/d1/bueHjTkCl4BYB5Tn0Myf/58TZkyRRkZGTr3nwcKVapUSZGRkUpISNDDDz98WevyHBLg0ngOCXCxK/EckhtGrHTLOvsmdXbLOuWRqbdBPPLII3rkkUdUXFys4/95hHKtWrXk6elpZlgAALgVLRvnysV9mZ6envYv8wEAANeecpGQAABQkVX0O2TcgYQEAACDkY84x3fZAAAA01EhAQDAYB4elEicISEBAMBgtGyco2UDAABMR4UEAACDcZeNcyQkAAAYjHzEORISAAAMRoXEOfaQAAAA01EhAQDAYFRInCMhAQDAYOQjztGyAQAApqNCAgCAwWjZOEdCAgCAwchHnKNlAwAATEeFBAAAg9GycY6EBAAAg5GPOEfLBgAAmI4KCQAABqNl4xwJCQAABiMfcY6EBAAAg1EhcY49JAAAwHRUSAAAMBgFEudISAAAMBgtG+do2QAAANNRIQEAwGAUSJwjIQEAwGC0bJyjZQMAAExHhQQAAINRIHGOCgkAAAazWCxuOVwxc+ZMRUREyNfXV76+voqKitKXX35pP3/27FnFx8crICBA1atXV1xcnHJzcx3WOHTokGJjY1W1alUFBgZqxIgRKikpcZiTmpqqli1bymq1KiwsTMnJyZf1MyIhAQCgAqpTp44mTJigjIwMffvtt7rrrrvUtWtX7dq1S5I0bNgwLVu2TJ9++qnS0tJ0+PBhPfDAA/b3nzt3TrGxsSoqKtLGjRs1Z84cJScna8yYMfY52dnZio2NVYcOHZSZmamhQ4fqySef1KpVq1yO12Kz2Wz/+2WXL943DzI7BKBc+mXrdLNDAMqdKldg80K7Nza4ZZ11Cbf/T+/39/fXpEmT9OCDD6p27dqaN2+eHnzwQUnS3r171aRJE6Wnp6tNmzb68ssv1aVLFx0+fFhBQUGSpFmzZmnUqFE6duyYvLy8NGrUKK1YsUI7d+60f0b37t116tQprVy50qXYqJAAAGAwi8U9R2FhofLy8hyOwsJCp59/7tw5ffLJJyooKFBUVJQyMjJUXFys6Oho+5zGjRurXr16Sk9PlySlp6erWbNm9mREkmJiYpSXl2evsqSnpzuscWHOhTVcQUICAIDB3LWHJCkpSX5+fg5HUlLSn37ujh07VL16dVmtVg0YMECLFi1SeHi4cnJy5OXlpRo1ajjMDwoKUk5OjiQpJyfHIRm5cP7Cub+ak5eXpzNnzrj0M+IuGwAArhKJiYlKSEhwGLNarX86v1GjRsrMzNTp06f12WefqXfv3kpLSzM6zMtCQgIAgMHcdduv1Wr9ywTkj7y8vBQWFiZJioyM1NatWzV16lQ98sgjKioq0qlTpxyqJLm5uQoODpYkBQcHa8uWLQ7rXbgL5/dz/nhnTm5urnx9feXt7e3StdGyAQDAYGbc9nsppaWlKiwsVGRkpDw9PbVmzRr7uaysLB06dEhRUVGSpKioKO3YsUNHjx61z0lJSZGvr6/Cw8Ptc36/xoU5F9ZwBRUSAAAqoMTERN1zzz2qV6+efv31V82bN0+pqalatWqV/Pz81LdvXyUkJMjf31++vr4aPHiwoqKi1KZNG0lSp06dFB4erp49e2rixInKycnR6NGjFR8fb6/SDBgwQNOnT9fIkSPVp08frV27VgsWLNCKFStcjpeEBAAAg5nxpNajR4+qV69eOnLkiPz8/BQREaFVq1bp7rvvliRNmTJFHh4eiouLU2FhoWJiYvT222/b31+pUiUtX75cAwcOVFRUlKpVq6bevXtr3Lhx9jkNGjTQihUrNGzYME2dOlV16tTR7NmzFRMT43K8PIcEuIbwHBLgYlfiOSR3T9/klnVSBrVxyzrlEXtIAACA6WjZAABgML5czzkSEgAADOaOO2QqOhISAAAM5kE+4hR7SAAAgOmokAAAYDBaNs6RkAAAYDDyEedo2QAAANO5nJBs27ZNO3bssL9esmSJunXrpueff15FRUVuDQ4AgIrA4qZ/KjKXE5KnnnpK33//vSTphx9+UPfu3VW1alV9+umnGjlypNsDBADgaudhcc9RkbmckHz//fdq0aKFJOnTTz9Vu3btNG/ePCUnJ+vzzz93d3wAAOAa4PKmVpvNptLSUknS6tWr1aVLF0lS3bp1dfz4cfdGBwBABcBdNs65nJC0atVKr776qqKjo5WWlqaZM2dKkrKzsxUUFOT2AAEAuNqRjzjncsvmzTff1LZt2zRo0CC98MILCgsLkyR99tlnuu2229weIAAAqPhcrpBEREQ43GVzwaRJk1SpUiW3BAUAQEXiQYnEqct+MFpGRob27NkjSQoPD1fLli3dFhQAABUJ+YhzLickR48e1SOPPKK0tDTVqFFDknTq1Cl16NBBn3zyiWrXru3uGAEAuKqxqdU5l/eQDB48WPn5+dq1a5dOnjypkydPaufOncrLy9OQIUOMiBEAAFRwLldIVq5cqdWrV6tJkyb2sfDwcM2YMUOdOnVya3AAAFQEFEicczkhKS0tlaen50Xjnp6e9ueTAACA/2JTq3Mut2zuuusuPfPMMzp8+LB97Oeff9awYcPUsWNHtwYHAACuDS4nJNOnT1deXp6uu+46NWzYUA0bNlSDBg2Ul5ent956y4gYAQC4qlncdFRkLrds6tatq23btmn16tXau3evJKlJkyaKjo52e3AAAFQE3GXjnEsJSXFxsby9vZWZmam7775bd999t1FxAQCAa4hLCYmnp6fq1aunc+fOGRUPAAAVjgcFEqdc3kPywgsv6Pnnn9fJkyeNiAcAgArHYrG45ajIXN5DMn36dO3fv1+hoaGqX7++qlWr5nB+27ZtbgsOAABcG1xOSLp162ZAGAAAVFwVvLjhFi4nJC+99JIRcQAAUGFV9HaLO7i8h0Q6/2V6s2fPVmJion0vybZt2/Tzzz+7NTgAACoCD4t7jorM5QrJ9u3bFR0dLT8/Px08eFD9+vWTv7+/Fi5cqEOHDunDDz80Ik4AAFCBuVwhSUhI0OOPP659+/apSpUq9vF7771X69atc2twAABUBNxl45zLFZKtW7fqnXfeuWj8b3/7m3JyctwSFAAAFUnFTiXcw+UKidVqVV5e3kXj33//vWrXru2WoAAAwLXF5YTk/vvv17hx41RcXCzpfBnq0KFDGjVqlOLi4tweIAAAVzsPi8UtR0XmckIyefJk5efnKzAwUGfOnNGdd96psLAw+fj4aPz48UbECADAVc1icc9Rkbm8h8TPz08pKSlav369tm/frvz8fLVs2ZJv+wUAAJfN5YTkgjvuuEN33HGHO2MBAKBCquh3yLhDmRKSadOmlXnBIUOGXHYwAABUROQjzpUpIZkyZYrD62PHjum3335TjRo1JJ1/cmvVqlUVGBhIQgIAAFxWpk2t2dnZ9mP8+PFq0aKF9uzZo5MnT+rkyZPas2ePWrZsqVdeecXoeAEAuOqYcZdNUlKSbrnlFvn4+CgwMFDdunVTVlaWw5z27dtf9PC1AQMGOMw5dOiQYmNj7YWHESNGqKSkxGFOamqqWrZsKavVqrCwMCUnJ7v+M3L1DS+++KLeeustNWrUyD7WqFEjTZkyRaNHj3Y5AAAAKjoz7rJJS0tTfHy8Nm3apJSUFBUXF6tTp04qKChwmNevXz8dOXLEfkycONF+7ty5c4qNjVVRUZE2btyoOXPmKDk5WWPGjLHPyc7OVmxsrDp06KDMzEwNHTpUTz75pFatWuVSvC5vaj1y5MhFmdGFoHNzc11dDgCACs+MTa0rV650eJ2cnKzAwEBlZGSoXbt29vGqVasqODj4kmt89dVX2r17t1avXq2goCC1aNFCr7zyikaNGqWXX35ZXl5emjVrlho0aKDJkydLkpo0aaL169drypQpiomJKXO8LldIOnbsqKeeekrbtm2zj2VkZGjgwIHc+gsAgIEKCwuVl5fncBQWFpbpvadPn5Yk+fv7O4zPnTtXtWrVUtOmTZWYmKjffvvNfi49PV3NmjVTUFCQfSwmJkZ5eXnatWuXfc4f//8fExOj9PR0l67N5QrJ+++/r969e6tVq1by9PSUJJWUlCgmJkazZ892dTlDHNv0ltkhAOVSzdueNTsEoNw5s+V1wz/D5b/9/4mkpCSNHTvWYeyll17Syy+//JfvKy0t1dChQ3X77beradOm9vFHH31U9evXV2hoqLZv365Ro0YpKytLCxculCTl5OQ4JCOS7K8vfH/dn83Jy8vTmTNn5O3tXaZrczkhqV27tr744gt9//332rt3rySpcePGuvHGG11dCgCAa4K7WjaJiYlKSEhwGLNarU7fFx8fr507d2r9+vUO4/3797f/ulmzZgoJCVHHjh114MABNWzY0C0xl9VlPxjtxhtvJAkBAOAKslqtZUpAfm/QoEFavny51q1bpzp16vzl3NatW0uS9u/fr4YNGyo4OFhbtmxxmHNhv+iFfSfBwcEX7SHNzc2Vr69vmasj0mUmJD/99JOWLl2qQ4cOqaioyOHcG2+8cTlLAgBQYXmY8GA0m82mwYMHa9GiRUpNTVWDBg2cviczM1OSFBISIkmKiorS+PHjdfToUQUGBkqSUlJS5Ovrq/DwcPucL774wmGdlJQURUVFuRSvywnJmjVrdP/99+v666/X3r171bRpUx08eFA2m00tW7Z0dTkAACo8MxKS+Ph4zZs3T0uWLJGPj499z4efn5+8vb114MABzZs3T/fee68CAgK0fft2DRs2TO3atVNERIQkqVOnTgoPD1fPnj01ceJE5eTkaPTo0YqPj7dXagYMGKDp06dr5MiR6tOnj9auXasFCxZoxYoVLsXr8j6bxMREPfvss9qxY4eqVKmizz//XD/++KPuvPNOPfTQQ64uBwAADDBz5kydPn1a7du3V0hIiP2YP3++JMnLy0urV69Wp06d1LhxYw0fPlxxcXFatmyZfY1KlSpp+fLlqlSpkqKiovTYY4+pV69eGjdunH1OgwYNtGLFCqWkpKh58+aaPHmyZs+e7dItv5JksdlsNlfe4OPjo8zMTDVs2FA1a9bU+vXrddNNN+m7775T165ddfDgQZcCMEJ+oUuXBFwzarcdYXYIQLlzJe6yGb4sy/mkMph8XyPnk65SLldIqlWrZt83EhISogMHDtjPHT9+3H2RAQBQQXhY3HNUZC7vIWnTpo3Wr1+vJk2a6N5779Xw4cO1Y8cOLVy4UG3atDEiRgAAUMG5nJC88cYbys/PlySNHTtW+fn5mj9/vm644QbusAEA4BJMeHL8VcflhOT666+3/7patWqaNWuWWwMCAKCicfWbeq9Fl/1gNAAAUDbuenR8RVamhKRmzZplfuztyZMn/6eAAADAtadMCcmbb75p//WJEyf06quvKiYmxv4UtvT0dK1atUovvviiIUECAHA1o2PjnMvPIYmLi1OHDh00aNAgh/Hp06dr9erVWrx4sTvjuyw8hwS4NJ5DAlzsSjyH5MWV+9yyziudb3DLOuWRy22tVatWqXPnzheNd+7cWatXr3ZLUAAA4NrickISEBCgJUuWXDS+ZMkSBQQEuCUoAAAqEovFPUdF5vJdNmPHjtWTTz6p1NRU+9cUb968WStXrtS7777r9gABALjaVfSnrLqDywnJ448/riZNmmjatGlauHChJKlJkyZav369PUEBAABwhUsJSXFxsZ566im9+OKLmjt3rlExAQBQofBgNOdc2kPi6empzz//3KhYAACokNhD4pzLm1q7detWLm7tBQAAFYfLe0huuOEGjRs3Ths2bFBkZKSqVavmcH7IkCFuCw4AgIqATa3OuZyQvPfee6pRo4YyMjKUkZHhcM5isZCQAADwBxaRkTjjckKSnZ1tRBwAAFRYVEicu+wvICwqKlJWVpZKSkrcGQ8AALgGuZyQ/Pbbb+rbt6+qVq2qm266SYcOHZIkDR48WBMmTHB7gAAAXO08LO45KjKXE5LExER99913Sk1NVZUqVezj0dHRmj9/vluDAwCgIrBYLG45KjKX95AsXrxY8+fPV5s2bRx+ODfddJMOHDjg1uAAAMC1weWE5NixYwoMDLxovKCgoMJnbwAAXI6K3m5xB5dbNq1atdKKFSvsry8kIbNnz1ZUVJT7IgMAoILgSa3OlblCsnPnTjVt2lRJSUnq3Lmzdu/ereLiYk2dOlW7d+/Wxo0blZaWZmSsAACggipzhSQiIkKtW7fW7t27tWHDBpWUlCgiIkJfffWVAgMDlZ6ersjISCNjBQDgquRhsbjlqMjKXCFJS0vTBx98oOHDh6u0tFRxcXF6/fXX1a5dOyPjAwDgqsceEufKXCFp27at3n//fR05ckRvvfWWDh48qPbt2+vGG2/UP/7xD+Xk5BgZJwAAqMBc3tRarVo1PfHEE0pLS9P333+vhx56SDNmzFC9evV0//33GxEjAABXNTa1Oufybb+/FxYWpueff17169dXYmKiw903AADgPA++XM+py05I1q1bp/fff1+ff/65PDw89PDDD6tv377ujA0AgAqholc33MGlhOTw4cNKTk5WcnKy9u/fr9tuu03Tpk3Tww8/rGrVqhkVIwAAqODKnJDcc889Wr16tWrVqqVevXqpT58+atSokZGxAQBQIXCXjXNlTkg8PT312WefqUuXLqpUqZKRMQEAUKFU9GeIuEOZE5KlS5caGQcAALiG/U932QAAAOcokDhHQgIAgMFo2Tjn8oPRAAAA3I0KCQAABqNA4hwJCQAABqMd4Rw/IwAAYDoSEgAADGaxWNxyuCIpKUm33HKLfHx8FBgYqG7duikrK8thztmzZxUfH6+AgABVr15dcXFxys3NdZhz6NAhxcbGqmrVqgoMDNSIESNUUlLiMCc1NVUtW7aU1WpVWFiYkpOTXf4ZkZAAAGAwi5sOV6SlpSk+Pl6bNm1SSkqKiouL1alTJxUUFNjnDBs2TMuWLdOnn36qtLQ0HT58WA888ID9/Llz5xQbG6uioiJt3LhRc+bMUXJyssaMGWOfk52drdjYWHXo0EGZmZkaOnSonnzySa1atcq1n5HNZrO5eI3lXn5hhbskwC1qtx1hdghAuXNmy+uGf8a/Mn5yyzoPNa2twsJChzGr1Sqr1er0vceOHVNgYKDS0tLUrl07nT59WrVr19a8efP04IMPSpL27t2rJk2aKD09XW3atNGXX36pLl266PDhwwoKCpIkzZo1S6NGjdKxY8fk5eWlUaNGacWKFdq5c6f9s7p3765Tp05p5cqVZb42KiQAAFwlkpKS5Ofn53AkJSWV6b2nT5+WJPn7+0uSMjIyVFxcrOjoaPucxo0bq169ekpPT5ckpaenq1mzZvZkRJJiYmKUl5enXbt22ef8fo0Lcy6sUVbcZQMAgMHcdddvYmKiEhISHMbKUh0pLS3V0KFDdfvtt6tp06aSpJycHHl5ealGjRoOc4OCgpSTk2Of8/tk5ML5C+f+ak5eXp7OnDkjb2/vMl0bCQkAAAZz13NIytqe+aP4+Hjt3LlT69evd08gBqBlAwBABTZo0CAtX75cX3/9terUqWMfDw4OVlFRkU6dOuUwPzc3V8HBwfY5f7zr5sJrZ3N8fX3LXB2RSEgAADCcGbf92mw2DRo0SIsWLdLatWvVoEEDh/ORkZHy9PTUmjVr7GNZWVk6dOiQoqKiJElRUVHasWOHjh49ap+TkpIiX19fhYeH2+f8fo0Lcy6sUVa0bAAAMJgZf/uPj4/XvHnztGTJEvn4+Nj3fPj5+cnb21t+fn7q27evEhIS5O/vL19fXw0ePFhRUVFq06aNJKlTp04KDw9Xz549NXHiROXk5Gj06NGKj4+3t44GDBig6dOna+TIkerTp4/Wrl2rBQsWaMWKFS7FS4UEAIAKaObMmTp9+rTat2+vkJAQ+zF//nz7nClTpqhLly6Ki4tTu3btFBwcrIULF9rPV6pUScuXL1elSpUUFRWlxx57TL169dK4cePscxo0aKAVK1YoJSVFzZs31+TJkzV79mzFxMS4FC/PIQGuITyHBLjYlXgOyYLMw25Z5+EWoW5ZpzyiZQMAgMH4sl/naNkAAADTUSEBAMBgrt4hcy0iIQEAwGC0I5wjIQEAwGBUSJwjaQMAAKajQgIAgMGojzhHQgIAgMHo2DhHywYAAJiOCgkAAAbzoGnjFAkJAAAGo2XjHC0bAABgOiokAAAYzELLxikSEgAADEbLxjlaNgAAwHRUSAAAMBh32ThHQgIAgMFo2ThHQgIAgMFISJxjDwkAADAdFRIAAAzGbb/OkZAAAGAwD/IRp2jZAAAA01EhAQDAYLRsnCMhAQDAYNxl4xwtGwAAYDoqJAAAGIyWjXMkJAAAGIy7bJyjZQMAAExHhQQu6dL5Lh05fPii8YceeVTPvTBGx48f09Q3Jmlz+kYVFBSo/nUN1LffU+p4d4x97nv/nKX136QqK2uvPD09lbZh6xW8AsD9nu3VQa8MitX0j9dpxJSlqunrrRf7x6hj6xtVN6imjp/K17K0nRo7a5XyCs7a3zd5eFe1iWigmxoGa+/BXLV5bMpFa8dFN9eIx+/SDfVq6/gvBZr16QZN+VfqFbw6uAMtG+dISOCSj+Z9pnOl5+yvD+zfp6f791F0p/MJx5gXRin/11/1xrS3VaNmTa38YrmeGzFMH338mRo3CZckFRcXKbpTZzVr3kJLFn1uynUA7hLZpK76PhCl7fv+m6iH1PJTSC1fJU5drj3ZuaoXUlNvPRenkFp+ejTxQ4f3f7hsi25pWk9Nw0IuWrtTVGN9MO5RJby+WKs3ZalxgyC9/fyDOlNYrFmfbjD82uA+3GXjHC0buKSmv79q1aptP75JS1WduvUU2epWSdL2zEw98n+PqWmzCNWpU1dP9h8oHx8f7dm9y77GgPgh6tHzcYXdcKNZlwG4RTVvL33wyqN6evynOpV3xj6++4cc/d9zH+qL9buV/fMJpX27Xy/P/FL3tg1XpUr//WN3+OQleuezjcr++eQl13/03pZalrZTsxem6+Dhk1q5YY8mzVmr4b06GH5tcC+Lm46KjIQEl624uEhfrFiqrt0ekOU/6X9Eixb6atUXOn36lEpLS7XqyxUqLCxSq1tuNTlawP3eHPmAVm7Yo6+37nM617d6FeUVnNW5c6VlXt/qWVlnC0scxs4UFqtOUA3VC6npcrxAeXbVJySFhYXKy8tzOAoLC80O65rw9do1yv/1V93X9e/2sX9MelMlJSW6q20btWkVofGvvKTX33xLdevVNzFSwP0euruFWjT6m16c8YXTuQF+VZXY5269v3iTS5+RsilLXTs0U/tbwmSxWBRWr5aeefROSVJILd/Lihvm8LBY3HJUZOU6Ifnxxx/Vp0+fv5yTlJQkPz8/h2PyxKQrFOG1bcmiz3Tb7W1VOzDIPjZzxlT9mverZv7zA/3r48/0WM/H9dyIYdr3fZaJkQLuVSfQT5MSuuqJMfNUWFTyl3N9qlm1aMqT2pOdq1f/+ZVLn/P+4s2a9ekGLZzcV3kbJijtvSH6NCVTklRaarvc8GECWjbOletNrSdPntScOXP0/vvv/+mcxMREJSQkOIwVy8vo0K55Rw7/rC2b0jVpylv2sR9/PKT5H8/VgoXL1DDsBknSjY0a69/bMvTp/Hl6/sWxZoULuNXNTeooKMBH6R8OtY9VrlxJd9zcQAMeul1+dzyn0lKbqle1aunUfvr1t7N6ZGSySlxo11wwevoKjXn7CwUH+OjYLwXqcMv531vZP59w1+UA5YKpCcnSpUv/8vwPP/zgdA2r1Sqr1eowll/I3xyMtnTxQtX0D9Adbe+0j509c35Tn4eHY+HNo5KHSktd/4MYKK++3rpfkd1fdxj755hHlHXwqCZ/+LVKS23yqWbVsmn9VFh0Tg8O/8BpJeWvlJbadPhYniTp4ZgW2rT9oI6fKvifrgFXWEUvb7iBqQlJt27dZLFYZLP9eQJhqeA9s6tRaWmpli5ZpC73d1Plyv/9T+i6Bterbr36Gj/uJQ0dPlJ+NWoode1qbU7fqDenz7LPO3LksPJOn1bOkSMqPXdOWXv3SJLq1qunqlWrXfHrAVyV/1uhdv+Q4zBWcKZIJ08XaPcPOfKpZtXyaf3lXcVTT4yZI9/qVeRbvYok6dgv+fZ2y/V1AlTd26qgAB95Wz0VcUOoJGlPdq6KS84pwK+q/t6xudZl7FcVL0/1uu8WPXBXc3Ua8PaVvWD8z3gOiXOmJiQhISF6++231bVr10uez8zMVGRk5BWOCs5s3rRROUcOq2u3BxzGPT09NW3GO3rrzckaNnigfvvtN9WtV09jX53gUEmZNWOali9dbH/96MPnN8W+894ctbql9RW5BsBILRrV0a3Nzm/k3r0o0eFco67jdejIL5KkmS88rHaRDe3nNs9NuGjOY7GRShrSRRaLRZt3HFTMwJn6dvePV+IygCvKYvur8oTB7r//frVo0ULjxo275PnvvvtON998s8vlflo2wKXVbjvC7BCAcufMltedT/ofbfnhtFvWufV6P7esUx6ZWiEZMWKECgr+vA8aFhamr7/++gpGBACA+9Gwcc7U237btm2rzp07/+n5atWq6c477/zT8wAA4M+tW7dO9913n0JDQ2WxWLR48WKH848//rgsFovD8cf/L588eVI9evSQr6+vatSoob59+yo/P99hzvbt29W2bVtVqVJFdevW1cSJE12OtVw/hwQAgArBpAeRFBQUqHnz5poxY8afzuncubOOHDliPz7++GOH8z169NCuXbuUkpKi5cuXa926derfv7/9fF5enjp16qT69esrIyNDkyZN0ssvv6x//vOfLsVarp9DAgBARWDWXTb33HOP7rnnnr+cY7VaFRwcfMlze/bs0cqVK7V161a1atVKkvTWW2/p3nvv1euvv67Q0FDNnTtXRUVFev/99+Xl5aWbbrpJmZmZeuONNxwSF2eokAAAYDCLxT2HEV+XkpqaqsDAQDVq1EgDBw7UiRP/feheenq6atSoYU9GJCk6OloeHh7avHmzfU67du3k5fXfh5LGxMQoKytLv/zyS5njICEBAOAqcamvS0lKuvyvS+ncubM+/PBDrVmzRv/4xz+Ulpame+65R+fOnZMk5eTkKDAw0OE9lStXlr+/v3JycuxzgoKCHOZceH1hTlnQsgEAwGDuathc6utS/vi0cld0797d/utmzZopIiJCDRs2VGpqqjp27HjZ614OKiQAABjNTZtarVarfH19HY7/JSH5o+uvv161atXS/v37JUnBwcE6evSow5ySkhKdPHnSvu8kODhYubm5DnMuvP6zvSmXQkICAAAkST/99JNOnDihkJAQSVJUVJROnTqljIwM+5y1a9eqtLRUrVu3ts9Zt26diouL7XNSUlLUqFEj1axZs8yfTUICAIDBLG76x1X5+fnKzMxUZmamJCk7O1uZmZk6dOiQ8vPzNWLECG3atEkHDx7UmjVr1LVrV4WFhSkmJkaS1KRJE3Xu3Fn9+vXTli1btGHDBg0aNEjdu3dXaOj571569NFH5eXlpb59+2rXrl2aP3++pk6delFryRn2kAAAYDCzvif222+/VYcOHeyvLyQJvXv31syZM7V9+3bNmTNHp06dUmhoqDp16qRXXnnFoQ00d+5cDRo0SB07dpSHh4fi4uI0bdo0+3k/Pz999dVXio+PV2RkpGrVqqUxY8a4dMuvZPJ32RiF77IBLo3vsgEudiW+yybz0K9uWadFPR+3rFMeUSEBAMBgfJeNcyQkAAAYjYzEKTa1AgAA01EhAQDAYGZ9l83VhIQEAACDmXWXzdWEhAQAAIORjzjHHhIAAGA6KiQAABiNEolTJCQAABiMTa3O0bIBAACmo0ICAIDBuMvGORISAAAMRj7iHC0bAABgOiokAAAYjRKJUyQkAAAYjLtsnKNlAwAATEeFBAAAg3GXjXMkJAAAGIx8xDkSEgAAjEZG4hR7SAAAgOmokAAAYDDusnGOhAQAAIOxqdU5WjYAAMB0VEgAADAYBRLnSEgAADAaGYlTtGwAAIDpqJAAAGAw7rJxjoQEAACDcZeNc7RsAACA6aiQAABgMAokzpGQAABgNDISp0hIAAAwGJtanWMPCQAAMB0VEgAADMZdNs6RkAAAYDDyEedo2QAAANNRIQEAwGC0bJwjIQEAwHBkJM7QsgEAAKajQgIAgMFo2ThHhQQAAINZ3HS4at26dbrvvvsUGhoqi8WixYsXO5y32WwaM2aMQkJC5O3trejoaO3bt89hzsmTJ9WjRw/5+vqqRo0a6tu3r/Lz8x3mbN++XW3btlWVKlVUt25dTZw40eVYSUgAAKigCgoK1Lx5c82YMeOS5ydOnKhp06Zp1qxZ2rx5s6pVq6aYmBidPXvWPqdHjx7atWuXUlJStHz5cq1bt079+/e3n8/Ly1OnTp1Uv359ZWRkaNKkSXr55Zf1z3/+06VYLTabzXZ5l1l+5RdWuEsC3KJ22xFmhwCUO2e2vG74Zxw5XeSWdUL8vC77vRaLRYsWLVK3bt0kna+OhIaGavjw4Xr22WclSadPn1ZQUJCSk5PVvXt37dmzR+Hh4dq6datatWolSVq5cqXuvfde/fTTTwoNDdXMmTP1wgsvKCcnR15e5+N77rnntHjxYu3du7fM8VEhAQDAYBY3/VNYWKi8vDyHo7Cw8LJiys7OVk5OjqKjo+1jfn5+at26tdLT0yVJ6enpqlGjhj0ZkaTo6Gh5eHho8+bN9jnt2rWzJyOSFBMTo6ysLP3yyy9ljoeEBAAAo7lpE0lSUpL8/PwcjqSkpMsKKScnR5IUFBTkMB4UFGQ/l5OTo8DAQIfzlStXlr+/v8OcS63x+88oC+6yAQDgKpGYmKiEhASHMavValI07kVCAgCAwdx116/VanVbAhIcHCxJys3NVUhIiH08NzdXLVq0sM85evSow/tKSkp08uRJ+/uDg4OVm5vrMOfC6wtzyoKWDQAABrNY3HO4U4MGDRQcHKw1a9bYx/Ly8rR582ZFRUVJkqKionTq1CllZGTY56xdu1alpaVq3bq1fc66detUXFxsn5OSkqJGjRqpZs2aZY6HhAQAgAoqPz9fmZmZyszMlHR+I2tmZqYOHToki8WioUOH6tVXX9XSpUu1Y8cO9erVS6GhofY7cZo0aaLOnTurX79+2rJlizZs2KBBgwape/fuCg0NlSQ9+uij8vLyUt++fbVr1y7Nnz9fU6dOvai15AwtGwAADGYx6btsvv32W3Xo0MH++kKS0Lt3byUnJ2vkyJEqKChQ//79derUKd1xxx1auXKlqlSpYn/P3LlzNWjQIHXs2FEeHh6Ki4vTtGnT7Of9/Pz01VdfKT4+XpGRkapVq5bGjBnj8KySsuA5JMA1hOeQABe7Es8hOZZf4pZ1alevuHUEWjYAAMB0FTfVAgCgnOC79ZwjIQEAwGB8269ztGwAAIDpqJAAAGAws+6yuZqQkAAAYDBaNs7RsgEAAKYjIQEAAKajZQMAgMFo2ThHQgIAgMHY1OocLRsAAGA6KiQAABiMlo1zJCQAABiMfMQ5WjYAAMB0VEgAADAaJRKnSEgAADAYd9k4R8sGAACYjgoJAAAG4y4b50hIAAAwGPmIcyQkAAAYjYzEKfaQAAAA01EhAQDAYNxl4xwJCQAABmNTq3O0bAAAgOksNpvNZnYQqJgKCwuVlJSkxMREWa1Ws8MByg1+bwAXIyGBYfLy8uTn56fTp0/L19fX7HCAcoPfG8DFaNkAAADTkZAAAADTkZAAAADTkZDAMFarVS+99BKb9oA/4PcGcDE2tQIAANNRIQEAAKYjIQEAAKYjIQEAAKYjIQEAAKYjIYFhZsyYoeuuu05VqlRR69attWXLFrNDAky1bt063XfffQoNDZXFYtHixYvNDgkoN0hIYIj58+crISFBL730krZt26bmzZsrJiZGR48eNTs0wDQFBQVq3ry5ZsyYYXYoQLnDbb8wROvWrXXLLbdo+vTpkqTS0lLVrVtXgwcP1nPPPWdydID5LBaLFi1apG7dupkdClAuUCGB2xUVFSkjI0PR0dH2MQ8PD0VHRys9Pd3EyAAA5RUJCdzu+PHjOnfunIKCghzGg4KClJOTY1JUAIDyjIQEAACYjoQEblerVi1VqlRJubm5DuO5ubkKDg42KSoAQHlGQgK38/LyUmRkpNasWWMfKy0t1Zo1axQVFWViZACA8qqy2QGgYkpISFDv3r3VqlUr3XrrrXrzzTdVUFCgJ554wuzQANPk5+dr//799tfZ2dnKzMyUv7+/6tWrZ2JkgPm47ReGmT59uiZNmqScnBy1aNFC06ZNU+vWrc0OCzBNamqqOnTocNF47969lZycfOUDAsoREhIAAGA69pAAAADTkZAAAADTkZAAAADTkZAAAADTkZAAAADTkZAAAADTkZAAAADTkZAAAADTkZAAV5HHH39c3bp1s79u3769hg4dasjaAHAl8V02gBs8/vjjmjNnjiTJ09NT9erVU69evfT888+rcmXjfpstXLhQnp6ebllr6tSp4sHNAMxCQgK4SefOnfXBBx+osLBQX3zxheLj4+Xp6anExESHeUVFRfLy8nLLZ/r7+7tlHUny8/Nz21oA4CpaNoCbWK1WBQcHq379+ho4cKCio6O1dOlSeytk/PjxCg0NVaNGjSRJP/74ox5++GHVqFFD/v7+6tq1qw4ePGhf79y5c0pISFCNGjUUEBCgkSNHXlTB+GPLprCwUKNGjVLdunVltVoVFham9957z35+165d6tKli3x9feXj46O2bdvqwIEDki5u2RQWFmrIkCEKDAxUlSpVdMcdd2jr1q3286mpqbJYLFqzZo1atWqlqlWr6rbbblNWVpZDjEuWLFHLli1VpUoVXX/99Ro7dqxKSkokSTabTS+//LLq1asnq9Wq0NBQDRky5H/69wDg6kRCAhjE29tbRUVFkqQ1a9YoKytLKSkpWr58uYqLixUTEyMfHx9988032rBhg6pXr67OnTvb3zN58mQlJyfr/fff1/r163Xy5EktWrToLz+zV69e+vjjjzVt2jTt2bNH77zzjqpXry5J+vnnn9WuXTtZrVatXbtWGRkZ6tOnjz05+KORI0fq888/15w5c7Rt2zaFhYUpJiZGJ0+edJj3wgsvaPLkyfr2229VuXJl9enTx37um2++Ua9evfTMM89o9+7deuedd5ScnKzx48dLkj7//HNNmTJF77zzjvbt26fFixerWbNml/cDB3B1swH4n/Xu3dvWtWtXm81ms5WWltpSUlJsVqvV9uyzz9p69+5tCwoKshUWFtrnf/TRR7ZGjRrZSktL7WOFhYU2b29v26pVq2w2m80WEhJimzhxov18cXGxrU6dOvbPsdlstjvvvNP2zDPP2Gw2my0rK8smyZaSknLJGBMTE20NGjSwFRUVOb2G/Px8m6enp23u3Ln280VFRbbQ0FB7TF9//bVNkm316tX2OStWrLBJsp05c8Zms9lsHTt2tL322msOn/PRRx/ZQkJCbDabzTZ58mTbjTfe+KcxAbh2UCEB3GT58uWqXr26qlSponvuuUePPPKIXn75ZUlSs2bNHPaNfPfdd9q/f798fHxUvXp1Va9eXf7+/jp79qwOHDig06dP68iRI2rdurX9PZUrV1arVq3+9PMzMzNVqVIl3XnnnX96vm3btmXaBHvgwAEVFxfr9ttvt495enrq1ltv1Z49exzmRkRE2H8dEhIiSTp69Kj9OseNG2e/xurVq6tfv346cuSIfvvtNz300EM6c+aMrr/+evXr10+LFi3604oNgIqNTa2Am3To0EEzZ86Ul5eXQkNDHe6uqVatmsPc/Px8RUZGau7cuRetU7t27cv6fG9v7//p/OX6fYJjsVgkSaWlpZLOX+fYsWP1wAMPXPS+KlWqqG7dusrKytLq1auVkpKip59+WpMmTVJaWprb7h4CcHWgQgK4SbVq1RQWFqZ69eo5vdW3ZcuW2rdvnwIDAxUWFuZw+Pn5yc/PTyEhIdq8ebP9PSUlJcrIyPjTNZs1a6bS0lKlpaVd8nxERIS++eYbFRcXO72Whg0bysvLSxs2bLCPFRcXa+vWrQoPD3f6/t9fZ1ZW1kXXGBYWJg+P83/8eHt767777tO0adOUmpqq9PR07dixo8yfAaBiICEBTNCjRw/VqlVLXbt21TfffKPs7GylpqZqyJAh+umnnyRJzzzzjCZMmKDFixdr7969evrpp3Xq1Kk/XfO6665T79691adPHy1evNi+5oIFCyRJgwYNUl5enrp3765vv/1W+/bt00cffXTRXTHS+eRq4MCBGjFihFauXKndu3erX79++u2339S3b98yX+eYMWP04YcfauzYsdq1a5f27NmjTz75RKNHj5YkJScn67333tPOnTv1ww8/6F//+pe8vb1Vv359F36aACoCEhLABFWrVtW6detUr149PfDAA2rSpIn69u2rs2fPytfXV5I0fPhw9ezZU71791ZUVJR8fHz097///S/XnTlzph588EE9/fTTaty4sfr166eCggJJUkBAgNauXav8/HzdeeedioyM1LvvvvunrZEJEyYoLi5OPXv2VMuWLbV//36tWrVKNWvWLPN1xsTEaPny5frqq690yy23qE2bNpoyZYo94ahRo4beffdd3X777YqIiNDq1au1bNkyBQQElPkzAFQMFpuNRzMCAABzUSEBAACmIyEBAACmIyEBAACmIyEBAACmIyEBAACmIyEBAACmIyEBAACmIyEBAACmIyEBAACmIyEBAACmIyEBAACm+/86pH9Ctz7sqwAAAABJRU5ErkJggg==\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "clf = MultinomialNB(alpha=1).fit(x_train_tfidf_lemma, y_train)\n",
        "y_pred = clf.predict(x_validation_tfidf_lemma)\n",
        "y_pred = map_sentiment(pd.Series(y_pred), invert=True)\n",
        "plot_confussion_matrix(y_validation, y_pred)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QYCPq2FrXSc0"
      },
      "source": [
        "## Mejor modelo"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bDqreA14ecsH"
      },
      "source": [
        "Luego de obsevar el rendimiento de cada uno de los modelos naive bayes en su validation dataset, decidimos quedarnos con el que utiliza **N-gramas con tokenizacion sin lemmas**, ya que tubo el **f1 score** mas alto, llegando a **0.87**."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "I6OTwJahYJ-f"
      },
      "source": [
        "### Refitting completo\n",
        "\n",
        "Re-entrenamos el modelo pero esta vez usando la totalidad del training set"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-bK9FA1we72q",
        "outputId": "802707e5-41a0-4c9b-f2dd-9424682ea03e"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<50000x189296 sparse matrix of type '<class 'numpy.int64'>'\n",
              "\twith 7169809 stored elements in Compressed Sparse Row format>"
            ]
          },
          "metadata": {},
          "execution_count": 77
        }
      ],
      "source": [
        "count_vect = CountVectorizer(preprocessor=clean_text,\n",
        "                             tokenizer=tokenizer,\n",
        "                             min_df=5,\n",
        "                             stop_words=stoplist_tokenized,\n",
        "                             ngram_range=(1,2))\n",
        "x_train = count_vect.fit_transform(x_train_text_completo)\n",
        "x_test = count_vect.transform(x_test_text)\n",
        "x_train"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "HDCXahARflGD"
      },
      "outputs": [],
      "source": [
        "naive_bayes = MultinomialNB(alpha=1).fit(x_train, y_train_completo)\n",
        "y_pred = naive_bayes.predict(x_test) # predigo en el set de testeo\n",
        "y_pred = map_sentiment(pd.Series(y_pred), invert=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WRUCQS54v0j0"
      },
      "source": [
        "El formato esperado para el resultado es"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "6CcpoQJZvz2s",
        "outputId": "1cca0f32-a650-4b04-aa12-8a5efa4e8935"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "      ID sentimiento\n",
              "0  60000    positivo\n",
              "1  60001    positivo\n",
              "2  60002    negativo\n",
              "3  60003    positivo\n",
              "4  60004    positivo"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-d1892818-aeae-462d-b2a4-77b2ac721b3e\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>ID</th>\n",
              "      <th>sentimiento</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>60000</td>\n",
              "      <td>positivo</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>60001</td>\n",
              "      <td>positivo</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>60002</td>\n",
              "      <td>negativo</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>60003</td>\n",
              "      <td>positivo</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>60004</td>\n",
              "      <td>positivo</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-d1892818-aeae-462d-b2a4-77b2ac721b3e')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-d1892818-aeae-462d-b2a4-77b2ac721b3e button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-d1892818-aeae-462d-b2a4-77b2ac721b3e');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-f9b9fc77-e881-4724-a99f-4731e458d1dd\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-f9b9fc77-e881-4724-a99f-4731e458d1dd')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-f9b9fc77-e881-4724-a99f-4731e458d1dd button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "\n",
              "    </div>\n",
              "  </div>\n"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "dataframe",
              "variable_name": "sample",
              "summary": "{\n  \"name\": \"sample\",\n  \"rows\": 19,\n  \"fields\": [\n    {\n      \"column\": \"ID\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 5,\n        \"min\": 60000,\n        \"max\": 60018,\n        \"num_unique_values\": 19,\n        \"samples\": [\n          60000,\n          60005,\n          60011\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"sentimiento\",\n      \"properties\": {\n        \"dtype\": \"category\",\n        \"num_unique_values\": 2,\n        \"samples\": [\n          \"negativo\",\n          \"positivo\"\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    }\n  ]\n}"
            }
          },
          "metadata": {},
          "execution_count": 79
        }
      ],
      "source": [
        "sample.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "irEl6y9mvhJp"
      },
      "outputs": [],
      "source": [
        "resultados_naive_bayes = build_results(y_pred)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ukR14Joxwt4L",
        "outputId": "59b8d252-5abc-4bac-da18-2b27be2780a7"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Positives: 4595\n",
            "Negatives: 4004\n"
          ]
        }
      ],
      "source": [
        "print_positive_negatives(resultados_naive_bayes['sentimiento'])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-e8qRltmvUlW"
      },
      "source": [
        "# Guardado de resultados y modelo"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1NC367hqvbMD",
        "outputId": "fbac97e6-f526-4de3-f433-08a18efafdde"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['/content/drive/MyDrive/7506R-1C2024-GRUPO02/TP2/Models/naive_bayes.pkl']"
            ]
          },
          "metadata": {},
          "execution_count": 82
        }
      ],
      "source": [
        "# naive bayes\n",
        "resultados_naive_bayes.to_csv(root_path + \"Dataset/resultados_naive_bayes.csv\", index=False)\n",
        "joblib.dump(naive_bayes, root_path + 'Models/naive_bayes.pkl')"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "provenance": [],
      "toc_visible": true,
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}