2.4-DecisionTree_MNist.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyMegDmpXNJCW4q1v8qsCwX6"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "#Modelo de Classificação - Decision Tree - Dataset MNist"
      ],
      "metadata": {
        "id": "pGNT810eLR7U"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "RyurLx8ATyhJ"
      },
      "outputs": [],
      "source": [
        "from sklearn.datasets import fetch_openml\n",
        "\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "\n",
        "import matplotlib\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn import tree\n",
        "\n",
        "from sklearn.metrics import confusion_matrix,ConfusionMatrixDisplay\n",
        "\n",
        "import time"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "mnist = fetch_openml('mnist_784')"
      ],
      "metadata": {
        "id": "_NVjMqdmUHnV"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "mnist"
      ],
      "metadata": {
        "id": "CmfygIrPUIof"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "X = mnist[\"data\"]\n",
        "y = mnist[\"target\"]\n",
        "\n",
        "print(X.shape)\n",
        "print(y.shape)"
      ],
      "metadata": {
        "id": "mFz3O4wZUIrF"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)"
      ],
      "metadata": {
        "id": "gBshYyUAUItt"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "inicio = time.time()\n",
        "modelo_dt = tree.DecisionTreeClassifier(criterion = 'entropy')\n",
        "modelo_dt.fit(X_train,y_train)\n",
        "termino = time.time()\n",
        "print(\"--- %s segundos para treinar o modelo ---\" % (termino - inicio))"
      ],
      "metadata": {
        "id": "m_vvMzrqUPqn"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "inicio = time.time()\n",
        "predicao = modelo_dt.predict(X_test)\n",
        "termino = time.time()\n",
        "print(predicao)\n",
        "print(\"--- %s segundos para fazer o scoring do modelo ---\" % (termino - inicio))"
      ],
      "metadata": {
        "id": "hdbsKD6iUPv1"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "matriz_confusao = confusion_matrix(y_test, predicao)\n",
        "print(matriz_confusao)\n",
        "\n",
        "matriz_confusao_visual = ConfusionMatrixDisplay(confusion_matrix=matriz_confusao,display_labels = np.array([1,2]))\n",
        "matriz_confusao_visual.plot();"
      ],
      "metadata": {
        "id": "NiY9mUbYUP1_"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize=(100,50))\n",
        "tree.plot_tree(modelo_dt, class_names=str(y_train.unique()) , filled=True, fontsize=10);\n",
        "plt.savefig('modelo_dt_mnist.png', dpi=300)"
      ],
      "metadata": {
        "id": "KlBUIZ_RUTGI"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}