StrokePrediction.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "StrokePrediction.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "authorship_tag": "ABX9TyPR+zokqVJ1SdhPYIWeBXgv",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/davidedomini/stroke_predictions/blob/main/StrokePrediction.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sJccl1lvU5jy"
      },
      "source": [
        "# Stroke Prediction\n",
        "\n",
        "**Davide Domini** <br> \n",
        "davide.domini@studio.unibo.it<br> <br>\n",
        "Programmazione di applicazioni data intensive <br>\n",
        "Laurea in Ingegneria e Scienze Informatiche  \n",
        "DISI - Università di Bologna, Cesena\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yyWprJtbWWE9"
      },
      "source": [
        "Citazioni: \n",
        "*   Stroke Preditction Dataset https://www.kaggle.com/fedesoriano/stroke-prediction-dataset\n",
        "\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "n5iHvB6fYZRM"
      },
      "source": [
        "##Descrizione del problema e comprensione dei dati"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "e12KrgYeYr8m"
      },
      "source": [
        "*In questo progetto si vuole realizzare un modello in grado di predire la presenza o meno di un ictus in base ad alcune caratteristiche fisiche e di stile di vita di alcuni pazienti*"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "h96nkZHrZhzN"
      },
      "source": [
        "Vengono importate le librerie necessarie\n",
        "- **NumPy** per lavorare agilmente con l'algebra lineare\n",
        "- **Pandas** per gestire meglio i dati in formato tabellare\n",
        "- **Seaborn** per disegnare i grafici (basata su **matplotlib**)\n",
        "- **Urllib** per recuperare il dataset dalla repo github\n",
        "- **Sklearn** per avere i modelli di classificazione\n",
        "- **Imblearn** per applicare l'oversampling alla classe meno numerosa"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YRpGEfA9XTbr"
      },
      "source": [
        "import numpy as np\n",
        "import pandas as pd \n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sb\n",
        "import os.path\n",
        "import math\n",
        "from urllib.request import urlretrieve\n",
        "from imblearn.over_sampling import SMOTE\n",
        "from sklearn.linear_model import Perceptron\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.pipeline import Pipeline\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.metrics import confusion_matrix\n",
        "from sklearn.metrics import precision_score, recall_score, f1_score\n",
        "from sklearn.model_selection import KFold, StratifiedKFold\n",
        "from sklearn.model_selection import GridSearchCV\n",
        "from sklearn.metrics import mean_squared_error\n",
        "from sklearn.linear_model import LogisticRegression\n",
        "from sklearn.svm import SVC\n",
        "from sklearn.tree import DecisionTreeClassifier\n",
        "import graphviz\n",
        "from sklearn import tree\n",
        "from sklearn.dummy import DummyClassifier\n",
        "from xgboost import XGBClassifier\n",
        "%matplotlib inline"
      ],
      "execution_count": 96,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "U62onaFtaA_E"
      },
      "source": [
        "###Caricamento dei dati e preprocessing\n",
        "\n",
        "Viene recuperato il file del dataset dalla repo GitHub"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CDd6hs9rXvdy"
      },
      "source": [
        "dataset_url = \"https://raw.githubusercontent.com/davidedomini/stroke_predictions/main/healthcare-dataset-stroke-data.csv\"\n",
        "\n",
        "if not os.path.exists(\"healthcare-dataset-stroke-data.csv\"):\n",
        "    urlretrieve(dataset_url, \"healthcare-dataset-stroke-data.csv\")"
      ],
      "execution_count": 97,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 359
        },
        "id": "XvAK1w64age9",
        "outputId": "88e6cb11-d5ef-4b96-e648-cf2bc7a6bfbd"
      },
      "source": [
        "stroke_dataset = pd.read_csv(\"healthcare-dataset-stroke-data.csv\", sep=\",\")\n",
        "stroke_dataset.head(10)"
      ],
      "execution_count": 98,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>id</th>\n",
              "      <th>gender</th>\n",
              "      <th>age</th>\n",
              "      <th>hypertension</th>\n",
              "      <th>heart_disease</th>\n",
              "      <th>ever_married</th>\n",
              "      <th>work_type</th>\n",
              "      <th>Residence_type</th>\n",
              "      <th>avg_glucose_level</th>\n",
              "      <th>bmi</th>\n",
              "      <th>smoking_status</th>\n",
              "      <th>stroke</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>9046</td>\n",
              "      <td>Male</td>\n",
              "      <td>67.0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Private</td>\n",
              "      <td>Urban</td>\n",
              "      <td>228.69</td>\n",
              "      <td>36.6</td>\n",
              "      <td>formerly smoked</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>51676</td>\n",
              "      <td>Female</td>\n",
              "      <td>61.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Self-employed</td>\n",
              "      <td>Rural</td>\n",
              "      <td>202.21</td>\n",
              "      <td>NaN</td>\n",
              "      <td>never smoked</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>31112</td>\n",
              "      <td>Male</td>\n",
              "      <td>80.0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Private</td>\n",
              "      <td>Rural</td>\n",
              "      <td>105.92</td>\n",
              "      <td>32.5</td>\n",
              "      <td>never smoked</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>60182</td>\n",
              "      <td>Female</td>\n",
              "      <td>49.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Private</td>\n",
              "      <td>Urban</td>\n",
              "      <td>171.23</td>\n",
              "      <td>34.4</td>\n",
              "      <td>smokes</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>1665</td>\n",
              "      <td>Female</td>\n",
              "      <td>79.0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Self-employed</td>\n",
              "      <td>Rural</td>\n",
              "      <td>174.12</td>\n",
              "      <td>24.0</td>\n",
              "      <td>never smoked</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>56669</td>\n",
              "      <td>Male</td>\n",
              "      <td>81.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Private</td>\n",
              "      <td>Urban</td>\n",
              "      <td>186.21</td>\n",
              "      <td>29.0</td>\n",
              "      <td>formerly smoked</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>53882</td>\n",
              "      <td>Male</td>\n",
              "      <td>74.0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Private</td>\n",
              "      <td>Rural</td>\n",
              "      <td>70.09</td>\n",
              "      <td>27.4</td>\n",
              "      <td>never smoked</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>10434</td>\n",
              "      <td>Female</td>\n",
              "      <td>69.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>No</td>\n",
              "      <td>Private</td>\n",
              "      <td>Urban</td>\n",
              "      <td>94.39</td>\n",
              "      <td>22.8</td>\n",
              "      <td>never smoked</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>27419</td>\n",
              "      <td>Female</td>\n",
              "      <td>59.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Private</td>\n",
              "      <td>Rural</td>\n",
              "      <td>76.15</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Unknown</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>60491</td>\n",
              "      <td>Female</td>\n",
              "      <td>78.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Private</td>\n",
              "      <td>Urban</td>\n",
              "      <td>58.57</td>\n",
              "      <td>24.2</td>\n",
              "      <td>Unknown</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      id  gender   age  ...   bmi   smoking_status stroke\n",
              "0   9046    Male  67.0  ...  36.6  formerly smoked      1\n",
              "1  51676  Female  61.0  ...   NaN     never smoked      1\n",
              "2  31112    Male  80.0  ...  32.5     never smoked      1\n",
              "3  60182  Female  49.0  ...  34.4           smokes      1\n",
              "4   1665  Female  79.0  ...  24.0     never smoked      1\n",
              "5  56669    Male  81.0  ...  29.0  formerly smoked      1\n",
              "6  53882    Male  74.0  ...  27.4     never smoked      1\n",
              "7  10434  Female  69.0  ...  22.8     never smoked      1\n",
              "8  27419  Female  59.0  ...   NaN          Unknown      1\n",
              "9  60491  Female  78.0  ...  24.2          Unknown      1\n",
              "\n",
              "[10 rows x 12 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 98
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WoWkZXzraoOC"
      },
      "source": [
        "Osservando il dataframe notiamo che il campo `id` è solo un identificatore univoco del paziente, non avendo nessuna importanza ai fini del modello lo impostiamo come index"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 390
        },
        "id": "Y9MtX-Ula2d5",
        "outputId": "ed8c538b-6ae0-4a30-fba7-4b97a837ee8d"
      },
      "source": [
        "stroke_dataset.set_index(\"id\", inplace=True)\n",
        "stroke_dataset.head(10)"
      ],
      "execution_count": 99,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>gender</th>\n",
              "      <th>age</th>\n",
              "      <th>hypertension</th>\n",
              "      <th>heart_disease</th>\n",
              "      <th>ever_married</th>\n",
              "      <th>work_type</th>\n",
              "      <th>Residence_type</th>\n",
              "      <th>avg_glucose_level</th>\n",
              "      <th>bmi</th>\n",
              "      <th>smoking_status</th>\n",
              "      <th>stroke</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>id</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>9046</th>\n",
              "      <td>Male</td>\n",
              "      <td>67.0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Private</td>\n",
              "      <td>Urban</td>\n",
              "      <td>228.69</td>\n",
              "      <td>36.6</td>\n",
              "      <td>formerly smoked</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>51676</th>\n",
              "      <td>Female</td>\n",
              "      <td>61.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Self-employed</td>\n",
              "      <td>Rural</td>\n",
              "      <td>202.21</td>\n",
              "      <td>NaN</td>\n",
              "      <td>never smoked</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>31112</th>\n",
              "      <td>Male</td>\n",
              "      <td>80.0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Private</td>\n",
              "      <td>Rural</td>\n",
              "      <td>105.92</td>\n",
              "      <td>32.5</td>\n",
              "      <td>never smoked</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>60182</th>\n",
              "      <td>Female</td>\n",
              "      <td>49.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Private</td>\n",
              "      <td>Urban</td>\n",
              "      <td>171.23</td>\n",
              "      <td>34.4</td>\n",
              "      <td>smokes</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1665</th>\n",
              "      <td>Female</td>\n",
              "      <td>79.0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Self-employed</td>\n",
              "      <td>Rural</td>\n",
              "      <td>174.12</td>\n",
              "      <td>24.0</td>\n",
              "      <td>never smoked</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>56669</th>\n",
              "      <td>Male</td>\n",
              "      <td>81.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Private</td>\n",
              "      <td>Urban</td>\n",
              "      <td>186.21</td>\n",
              "      <td>29.0</td>\n",
              "      <td>formerly smoked</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>53882</th>\n",
              "      <td>Male</td>\n",
              "      <td>74.0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Private</td>\n",
              "      <td>Rural</td>\n",
              "      <td>70.09</td>\n",
              "      <td>27.4</td>\n",
              "      <td>never smoked</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10434</th>\n",
              "      <td>Female</td>\n",
              "      <td>69.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>No</td>\n",
              "      <td>Private</td>\n",
              "      <td>Urban</td>\n",
              "      <td>94.39</td>\n",
              "      <td>22.8</td>\n",
              "      <td>never smoked</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>27419</th>\n",
              "      <td>Female</td>\n",
              "      <td>59.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Private</td>\n",
              "      <td>Rural</td>\n",
              "      <td>76.15</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Unknown</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>60491</th>\n",
              "      <td>Female</td>\n",
              "      <td>78.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Private</td>\n",
              "      <td>Urban</td>\n",
              "      <td>58.57</td>\n",
              "      <td>24.2</td>\n",
              "      <td>Unknown</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       gender   age  hypertension  ...   bmi   smoking_status stroke\n",
              "id                                 ...                              \n",
              "9046     Male  67.0             0  ...  36.6  formerly smoked      1\n",
              "51676  Female  61.0             0  ...   NaN     never smoked      1\n",
              "31112    Male  80.0             0  ...  32.5     never smoked      1\n",
              "60182  Female  49.0             0  ...  34.4           smokes      1\n",
              "1665   Female  79.0             1  ...  24.0     never smoked      1\n",
              "56669    Male  81.0             0  ...  29.0  formerly smoked      1\n",
              "53882    Male  74.0             1  ...  27.4     never smoked      1\n",
              "10434  Female  69.0             0  ...  22.8     never smoked      1\n",
              "27419  Female  59.0             0  ...   NaN          Unknown      1\n",
              "60491  Female  78.0             0  ...  24.2          Unknown      1\n",
              "\n",
              "[10 rows x 11 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 99
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_pFF4QYlbsPI"
      },
      "source": [
        "###Comprensione delle feature"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BBjtADbWbwvK"
      },
      "source": [
        "Descrizione:\n",
        "\n",
        "1. **id**: identificatore univoco del paziente\n",
        "2. **gender**: [Nominale] genere del paziente, può essere \"Male\", \"Female\" o \"Other\"\n",
        "3. **age**: [Intervallo] età del paziente\n",
        "4. **hypertension**: [Nominale] rappresenta la presenza di ipertensione nel paziente, assume valore 0 se non presente e valore 1 se presente\n",
        "5. **heart_disease**: [Nominale] rappresenta la presenza di cardiopatia nel paziente, assume valore 0 se non presente e valore 1 se presente\n",
        "6. **ever_married**: [Nominale] indica se il paziente è mai stato sposato, assume valore \"Yes\" o \"No\"\n",
        "7. **work_type**: [Nominale] indica il tipo di lavoro del paziente, assume valori \"children\", \"Govt_jov\", \"Never_Worked\", \"Private\" o \"Self-employed\"\n",
        "8. **Residence_type**: [Nominale] indica la zona di residenza del paziente, assume valori \"Rural\" o \"Urban\"\n",
        "9. **avg_glucose_level**: [Intervallo] indica il livello di glucosio medio nel sangue\n",
        "10. **bmi**: [Intervallo] indica l'indice di massa corporea, si calcola come: $$ \\frac{massa}{altezza^2} $$ \n",
        "11. **smoking_status**: [Nominale] indica le abitudini del paziente con il fumo, assume valori \"formerly smoked\", \"never smoked\", \"smokes\" o \"Unknown\"\n",
        "12. **stroke**: [Nominale] indica se il paziente ha avuto un ictus, assume valore 0 oppure 1\n",
        "\n",
        "$\\Rightarrow$ Siccome la variabile `stroke` da predire è discreta si tratta di un problema di classificazione (con due classi)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uMzz0DT8l2oe"
      },
      "source": [
        "###Esplorazione delle singole feature"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 297
        },
        "id": "CqViM8SDhk-a",
        "outputId": "89a8c295-3bac-42a6-d012-29fc0a7e8675"
      },
      "source": [
        "stroke_dataset.describe()"
      ],
      "execution_count": 100,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>age</th>\n",
              "      <th>hypertension</th>\n",
              "      <th>heart_disease</th>\n",
              "      <th>avg_glucose_level</th>\n",
              "      <th>bmi</th>\n",
              "      <th>stroke</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>count</th>\n",
              "      <td>5110.000000</td>\n",
              "      <td>5110.000000</td>\n",
              "      <td>5110.000000</td>\n",
              "      <td>5110.000000</td>\n",
              "      <td>4909.000000</td>\n",
              "      <td>5110.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean</th>\n",
              "      <td>43.226614</td>\n",
              "      <td>0.097456</td>\n",
              "      <td>0.054012</td>\n",
              "      <td>106.147677</td>\n",
              "      <td>28.893237</td>\n",
              "      <td>0.048728</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>std</th>\n",
              "      <td>22.612647</td>\n",
              "      <td>0.296607</td>\n",
              "      <td>0.226063</td>\n",
              "      <td>45.283560</td>\n",
              "      <td>7.854067</td>\n",
              "      <td>0.215320</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>min</th>\n",
              "      <td>0.080000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>55.120000</td>\n",
              "      <td>10.300000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>25%</th>\n",
              "      <td>25.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>77.245000</td>\n",
              "      <td>23.500000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>50%</th>\n",
              "      <td>45.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>91.885000</td>\n",
              "      <td>28.100000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>75%</th>\n",
              "      <td>61.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>114.090000</td>\n",
              "      <td>33.100000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>max</th>\n",
              "      <td>82.000000</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>271.740000</td>\n",
              "      <td>97.600000</td>\n",
              "      <td>1.000000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "               age  hypertension  ...          bmi       stroke\n",
              "count  5110.000000   5110.000000  ...  4909.000000  5110.000000\n",
              "mean     43.226614      0.097456  ...    28.893237     0.048728\n",
              "std      22.612647      0.296607  ...     7.854067     0.215320\n",
              "min       0.080000      0.000000  ...    10.300000     0.000000\n",
              "25%      25.000000      0.000000  ...    23.500000     0.000000\n",
              "50%      45.000000      0.000000  ...    28.100000     0.000000\n",
              "75%      61.000000      0.000000  ...    33.100000     0.000000\n",
              "max      82.000000      1.000000  ...    97.600000     1.000000\n",
              "\n",
              "[8 rows x 6 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 100
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fjaaqMIIh5nG"
      },
      "source": [
        "Con il metodo describe possiamo ottenere varie informazioni sulle feature presenti nel dataset:\n",
        "\n",
        "\n",
        "* L'età media dei pazienti è circa 43 anni, il più giovane ha meno di un anno mentre il più anziano ne ha 82, inoltre possiamo notare che comunque il 50% ha più di 45 anni\n",
        "* Il livello di glucosio medio nel sangue è circa 106, di solito questo valore dovrebbe stare nell'intervallo [70;100], dal 75-esimo percentile possiamo notare che circa il 25% dei pazienti ha livelli preoccupanti che denotano una probabile presenza di diabete\n",
        "* La media dei valori del BMI assume valore 28, leggermente troppo alto in quanto una persona con peso regolare dovrebbe stare nell'intervallo [18;25], inoltre abbiamo 10 come valore minimo e 97 come valore massimo il che indica che abbiamo alcuni casi di grave magrezza e grave obesità\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NYIjJdUuesgg"
      },
      "source": [
        "Inoltre dal riassunto precedente del dataframe possiamo notare che ci sono alcuni valori NaN quindi controlliamo meglio"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "L9E-p9TXe2qa",
        "outputId": "3d7d7aed-d7a3-435d-8738-77faeb17c3ea"
      },
      "source": [
        "stroke_dataset.isna().sum()"
      ],
      "execution_count": 101,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "gender                 0\n",
              "age                    0\n",
              "hypertension           0\n",
              "heart_disease          0\n",
              "ever_married           0\n",
              "work_type              0\n",
              "Residence_type         0\n",
              "avg_glucose_level      0\n",
              "bmi                  201\n",
              "smoking_status         0\n",
              "stroke                 0\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 101
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gz4Q1NXJe9W3"
      },
      "source": [
        "Vista la presenza di 201 valori mancanti nella colonna bmi procediamo alla rimozione"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NY2NACvZfEtO",
        "outputId": "5603dd1a-e0d9-4dda-b66f-502538eb372a"
      },
      "source": [
        "stroke_dataset.dropna(inplace=True)\n",
        "stroke_dataset.isna().sum()"
      ],
      "execution_count": 102,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "gender               0\n",
              "age                  0\n",
              "hypertension         0\n",
              "heart_disease        0\n",
              "ever_married         0\n",
              "work_type            0\n",
              "Residence_type       0\n",
              "avg_glucose_level    0\n",
              "bmi                  0\n",
              "smoking_status       0\n",
              "stroke               0\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 102
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 248
        },
        "id": "T2QD-YSbm2Dq",
        "outputId": "abcba341-ca40-47af-b235-102f1ad5d04e"
      },
      "source": [
        "stroke_dataset[\"stroke\"].value_counts().plot.pie(autopct=\"%.1f%%\");"
      ],
      "execution_count": 103,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZdCW3Swzn5d7"
      },
      "source": [
        "Dal grafico a torta della feature `stroke` notiamo che le due classi sono estremamente sbilanciate, questo protrebbe creare problemi in seguito quindi nelle prossime sezioni verranno applicate tecniche come dare un peso diverso alle due classi oppure under/over -sampling di una delle due classi"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 369
        },
        "id": "-SFQNc28ULRY",
        "outputId": "6ebf148c-35e9-4c74-efa6-81bea194a7f3"
      },
      "source": [
        "sb.displot(stroke_dataset[\"avg_glucose_level\"]);"
      ],
      "execution_count": 104,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 360x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 369
        },
        "id": "no5fT6-kVYqL",
        "outputId": "f6723f42-a755-4b43-8f3e-d824342f2847"
      },
      "source": [
        "sb.displot(stroke_dataset[\"bmi\"]);"
      ],
      "execution_count": 105,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWEAAAFgCAYAAABqo8hyAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAZ10lEQVR4nO3df5BdZZ3n8fd3YvuLRALY08R0QphJ0DASAxUZGP2DAV3RchfdchjcGaUsnMxg3NFd1x1wq9Zxai20yoHRHTa7URhgZUEWcU1cF0VknXVrJUZlEHKHa0YwJCSdjghGdFKd8N0/7ulwvXYnN0mffm7f+35V3epznnPO7e/lJJ8cnnvO80RmIkkq49dKFyBJg8wQlqSCDGFJKsgQlqSCDGFJKuh5pQs4HhdffHHefffdpcuQpE7R7Y5z+kp47969pUuQpOMyp0NYkuY6Q1iSCjKEJakgQ1iSCjKEJakgQ1iSCqothCPihRGxOSL+LiIejoiPVO03RcSjEfFA9VpdtUdEfCoitkXEgxFxTl21SVKvqPNhjf3AhZn5s4gYAr4ZEf+r2vbBzLyzY/83Aiuq128D66ufktS3arsSzpafVatD1etwgxdfAtxSHfctYGFELKqrPknqBbX2CUfEvIh4ANgD3JOZ91ebPlp1OVwXES+o2hYDj7cdvqNqk6S+VWsIZ+bBzFwNjALnRsQrgauBVwCvBk4G/uxo3jMi1kbElojYMj4+PuM1S9JsmpW7IzLzKeA+4OLM3FV1OewH/gY4t9ptJ7Ck7bDRqq3zvTZk5prMXDM8PFx36ZJUq9q+mIuIYWAiM5+KiBcBrwc+HhGLMnNXRATwFuCh6pCNwHsj4nZaX8g9nZm76qqvH0xMTNBoNH6pbeXKlQwNDRWqSNLRqvPuiEXAzRExj9YV9x2Z+aWI+HoV0AE8APxJtf+XgTcB24CfA++qsba+0Gg0uPL6TSwYWQrAvrHtrF8Hq1atKlyZpG7VFsKZ+SBw9hTtF06zfwLr6qqnXy0YWcrC0eWly5B0jHxiTpIKMoQlqSBDWJIKMoQlqSBDWJIKMoQlqSBDWJIKMoQlqSBDWJIKqvOxZdWgfbyIZrNJ60FDSXOVITzHtI8XsXvrZk5cdlbpkiQdB7sj5qDJ8SJOOOXU0qVIOk6GsCQVZAhLUkGGsCQVZAhLUkHeHdFHnj14gGazeWjdqY6k3mcI95Fn9j7BNZv2M3zaL5zqSJojDOE+M3941OmOpDnEPmFJKsgQlqSCDGFJKsgQlqSCDGFJKsgQlqSCDGFJKsgQlqSCDGFJKsgQlqSCDGFJKsgQlqSCDGFJKsgQlqSCDGFJKqi2EI6IF0bE5oj4u4h4OCI+UrWfHhH3R8S2iPhcRDy/an9Btb6t2r6srtokqVfUeSW8H7gwM18FrAYujojzgI8D12XmcuAnwBXV/lcAP6nar6v2k6S+VlsIZ8vPqtWh6pXAhcCdVfvNwFuq5UuqdartF0VE1FWfJPWCWvuEI2JeRDwA7AHuAf4BeCozD1S77AAWV8uLgccBqu1PA6dM8Z5rI2JLRGwZHx+vs3xJql2tIZyZBzNzNTAKnAu8Ygbec0NmrsnMNcPDw8ddoySVNCt3R2TmU8B9wPnAwoiYnGB0FNhZLe8ElgBU208Efjwb9UlSKXXeHTEcEQur5RcBrwcatML4bdVulwNfrJY3VutU27+emVlXfZLUC+qc8n4RcHNEzKMV9ndk5pciYitwe0T8B+B7wA3V/jcA/zUitgFPApfVWJsk9YTaQjgzHwTOnqL9h7T6hzvb/xH4vbrqkaRe5BNzklSQISxJBRnCklSQISxJBRnCklSQISxJBRnCklSQISxJBRnCklRQnY8t6xhNTEzQaDQOra9cuZKhoaGCFUmqiyHcgxqNBldev4kFI0vZN7ad9etg1apVpcuSVANDuEctGFnKwtHlpcuQVDNDuMc9e/AAzWbz0Hqz2aSbET47jwO7NaReZAj3uGf2PsE1m/YzfNovANi9dTMnLjvrqI+zW0PqTYbwHDB/ePRQ18S+se3HdJyk3uQtapJUkCEsSQUZwpJUkCEsSQUZwpJUkCEsSQUZwpJUkCEsSQUZwpJUkCEsSQUZwpJUkCEsSQU5gE8P6JxJo9vhKiXNfYZwD2ifSQO6H67yaHSOL+zYwlJvMIR7RPtMGkczXGW32scXdmxhqXcYwgPE8YWl3uMXc5JUkCEsSQXVFsIRsSQi7ouIrRHxcES8r2r/84jYGREPVK83tR1zdURsi4hHIuINddUmSb2izj7hA8AHMvO7EbEA+E5E3FNtuy4zP9G+c0ScCVwG/BbwMuBrEXFGZh6ssUZJKqq2K+HM3JWZ362W9wENYPFhDrkEuD0z92fmo8A24Ny66pOkXjArfcIRsQw4G7i/anpvRDwYETdGxElV22Lg8bbDdjBFaEfE2ojYEhFbxsfHa6xakupXewhHxHzg88D7M/OnwHrgN4HVwC7gL4/m/TJzQ2auycw1w8PDM16vJM2mWkM4IoZoBfCtmXkXQGaOZebBzHwW+DTPdTnsBJa0HT5atUlS36rz7ogAbgAamXltW/uitt3eCjxULW8ELouIF0TE6cAKYHNd9UlSL6jz7ojXAO8Avh8RD1RtHwLeHhGrgQQeA/4YIDMfjog7gK207qxY550RkvpdbSGcmd8EYopNXz7MMR8FPlpXTZLUa3xiTpIKMoQlqSBDWJIKMoQlqSBDWJIKMoQlqSBDWJIKMoQlqSDnmBtAzrws9Q5DeAA587LUOwzhAeXMy1JvsE9YkgoyhCWpILsjCpmYmKDRaADQbDbJzMIVSSrBEC6k0Whw5fWbWDCylN1bN3PisrNKlySpALsjClowspSFo8s54ZRTS5ciqRBDWJIKMoQlqSBDWJIKMoQlqSBDWJIKMoQlqSBDWJIKMoQlqSBDWJIK8rHlAdc5wDs4yLs0mwzhAdc+wDvgIO/SLDOE5QDvUkH2CUtSQYawJBVkCEtSQYawJBVkCEtSQbWFcEQsiYj7ImJrRDwcEe+r2k+OiHsi4gfVz5Oq9oiIT0XEtoh4MCLOqas2SeoVdV4JHwA+kJlnAucB6yLiTOAq4N7MXAHcW60DvBFYUb3WAutrrE2SekJtIZyZuzLzu9XyPqABLAYuAW6udrsZeEu1fAlwS7Z8C1gYEYvqqk+SesGs9AlHxDLgbOB+YCQzd1WbdgMj1fJi4PG2w3ZUbZLUt2oP4YiYD3weeH9m/rR9W2YmkEf5fmsjYktEbBkfH5/BSiVp9tUawhExRCuAb83Mu6rmscluhurnnqp9J7Ck7fDRqu2XZOaGzFyTmWuGh4frK16SZkGdd0cEcAPQyMxr2zZtBC6vli8HvtjW/s7qLonzgKfbui0kqS91FcIR8Zpu2jq8BngHcGFEPFC93gR8DHh9RPwAeF21DvBl4IfANuDTwHu6+wiSNHd1O4rafwQ679udqu2QzPwmENNsvmiK/RNY12U9ktQXDhvCEXE+8DvAcET867ZNLwHm1VmYJA2CI10JPx+YX+23oK39p8Db6ipKkgbFYUM4M78BfCMibsrMH81STX1rYmKCRqMBQLPZpNUDI2mQddsn/IKI2AAsaz8mMy+so6h+1Wg0uPL6TSwYWcrurZs5cdlZpUuSVFi3Ifzfgf8MfAY4WF85/W/ByFIWji5n39j20qVI6gHdhvCBzHRAHUmaYd0+rLEpIt4TEYuqoShPjoiTa61MkgZAt1fCk0+4fbCtLYHfmNlyJGmwdBXCmXl63YVI0iDqKoQj4p1TtWfmLTNbjiQNlm67I17dtvxCWo8dfxcwhCXpOHTbHfEv29cjYiFwey0VSdIAOdahLJ8B7CeWpOPUbZ/wJp6bAWMesBK4o66iJGlQdNsn/Im25QPAjzJzRw31qLBnDx6g2WweWl+5ciVDQ0MFK5L6W7d9wt+IiBGe+4LuB/WVpJKe2fsE12zaz/Bpv2Df2HbWr4NVq1aVLkvqW93OrHEpsBn4PeBS4P6IcCjLPjV/eJSFo8tZMLK0dClS3+u2O+LfAa/OzD0AETEMfA24s67CJGkQdHt3xK9NBnDlx0dxrCRpGt1eCd8dEV8BbqvWf5/WxJySpONwpDnmlgMjmfnBiPjnwGurTf8PuLXu4iSp3x3pSvivgKsBMvMu4C6AiDir2vZPa61Okvrckfp1RzLz+52NVduyWiqSpAFypBBeeJhtL5rJQiRpEB0phLdExB91NkbEu4Hv1FOSJA2OI/UJvx/4QkT8Ac+F7hrg+cBb6yxMkgbBYUM4M8eA34mI3wVeWTX/z8z8eu2VSdIA6HbsiPuA+2qupe9MTEzQaDQOrTebTTLzMEdIGjTdPqyhY9BoNLjy+k2HxmDYvXUzJy47q3BVknqJIVyzBSNLWTi6HIB9Y9sLVyOp1zj+gyQVZAhLUkGGsCQVZJ+wptU51RE43ZE002oL4Yi4EXgzsCczX1m1/TnwR8B4tduHMvPL1bargSuAg8CfZuZX6qpN3Wmf6ghwuiOpBnVeCd8E/DVwS0f7dZnZPnEoEXEmcBnwW8DLgK9FxBmZebDG+tSFyamOJNWjtj7hzPxb4Mkud78EuD0z92fmo8A24Ny6apOkXlHii7n3RsSDEXFjRJxUtS0GHm/bZ0fV9isiYm1EbImILePj41PtIklzxmyH8HrgN4HVwC7gL4/2DTJzQ2auycw1w8PDM12fJM2qWQ3hzBzLzIOZ+SzwaZ7rctgJLGnbdbRqk6S+NqshHBGL2lbfCjxULW8ELouIF0TE6cAKYPNs1iZJJdR5i9ptwAXASyNiB/Bh4IKIWA0k8BjwxwCZ+XBE3AFsBQ4A67wzQtIgqC2EM/PtUzTfcJj9Pwp8tK56JKkX+diyJBVkCEtSQYawJBVkCEtSQYawJBVkCEtSQY4nPMPaZ1h2dmVJR2IIz7D2GZadXVnSkdgdUYPJGZZPOOXU0qVI6nGGsCQVZAhLUkGGsCQVZAhLUkGGsCQVZAhLUkGGsCQV5MMax6n9CTnwKTlJR8cQPk7tT8gBPiUn6agYwjNg8gk5gH1j2wtXI2kusU9YkgoyhCWpIENYkgoyhCWpIENYkgry7gh17dmDB2g2m4fWV65cydDQUMGKpLnPEFbXntn7BNds2s/wab9g39h21q+DVatWlS5LmtMMYR2V+cOjLBxd/itXxeCVsXQsDGEdk/arYsArY+kYGcI6ZpNXxZKOnXdHSFJBhrAkFWQIS1JBhrAkFVRbCEfEjRGxJyIeams7OSLuiYgfVD9PqtojIj4VEdsi4sGIOKeuuiSpl9R5JXwTcHFH21XAvZm5Ari3Wgd4I7Cieq0F1tdYlyT1jNpCODP/Fniyo/kS4OZq+WbgLW3tt2TLt4CFEbGortokqVfM9n3CI5m5q1reDYxUy4uBx9v221G17aJDRKyldbXM0qVL66v0MNrnlXNOOUnHo9jDGpmZEXHU6ZWZG4ANAGvWrCmSfu3zyjmnnKTjMdt3R4xNdjNUP/dU7TuBJW37jVZtPWtyXrkTTjm1dCmS5rDZDuGNwOXV8uXAF9va31ndJXEe8HRbt4Uk9a3auiMi4jbgAuClEbED+DDwMeCOiLgC+BFwabX7l4E3AduAnwPvqqsu1cOxhqVjU1sIZ+bbp9l00RT7JrCurlpUP8calo6No6hpxjiqmnT0fGxZkgoyhCWpIENYkgoyhCWpIENYkgry7gjNOGdilrpnCGvGOROz1D1DWLXwnmGpO/YJS1JBhrAkFWQIS1JBhrAkFWQIS1JBhrAkFWQIS1JBhrAkFWQIS1JBhrAkFWQIS1JBhrAkFWQIS1JBhrAkFWQIS1JBhrAkFWQIS1JBhrAkFWQIS1JBhrAkFWQIS1JBzrasWTUxMUGj0Ti0vnLlSoaGhgpWJJVlCKt2zx48QLPZBKDZbHLtVx/hJacuZd/Ydtavg1WrVhWuUCrHEFbtntn7BNds2s/wab9g99bNnLjsLBaOLi9dltQTioRwRDwG7AMOAgcyc01EnAx8DlgGPAZcmpk/KVGfZt784VEWji5n39j2Q23tV8iT7J7QoCl5Jfy7mbm3bf0q4N7M/FhEXFWt/1mZ0jQb2q+QAbsnNJB6qTviEuCCavlm4H9jCPe9yStkaVCVukUtga9GxHciYm3VNpKZu6rl3cDIVAdGxNqI2BIRW8bHx2ejVkmqTakr4ddm5s6I+HXgnoj4+/aNmZkRkVMdmJkbgA0Aa9asmXIfSZorilwJZ+bO6uce4AvAucBYRCwCqH7uKVGbJM2mWQ/hiDghIhZMLgP/BHgI2AhcXu12OfDF2a5NkmZbie6IEeALETH5+/9bZt4dEd8G7oiIK4AfAZcWqE2SZtWsh3Bm/hB41RTtPwYumu16JKmkXrpFrWd1jnfQbDbJ9DtBScfPEO5Co9Hgyus3sWBkKcChR28l6XgZwl1aMLL00EMF7Y/eStLxMITVMzrHknAcCQ0CQ1g9o30sCceR0KAwhNVTHEtCg8bpjSSpIENYkgqyO0I9yQHfNSgMYfUkB3zXoDCE1bP8kk6DwD5hSSrIEJakggxhSSrIPuFptI+c5qhpkupiCE+jfeQ0R02TVBe7Iw5jcuS0E045tXQpkvqUISxJBRnCklSQISxJBRnCklSQd0dozumceBUc3EdzlyGsOadz4tX2wX0MaM01hrDmpPaJV9sdLqClXmQIa85rH3u42Wwy/9eXOPqa5gxDWHNe+9jDPt2oucYQ1pzQebXbOZbH5NjD+8a2H9P7d/Yl24+s2WIIa06YiavdwwVte1/y07se5QNvaHLGGWdMua80kwxhzRnHe7XbHrRTfWE3+WXfvrHtXLPpQadW0qwwhDVQpruropNTK2m2GMLqa9P1JXfO5uyY0SrFEG7jQO79Z7q+5M7ZnL2rQqUYwm0cyL0/TdeX3N7lcLh+5var5omJCYBf+pJuqi/tfHJP3RroEO78i9J+o/+xfvmj/tN5NT3vhJMYPm0FMP2Xdj65p271XAhHxMXAJ4F5wGcy82Mz+f6dXQ7XfvURXnJq6y+KV7+aTvvV9PMWDE/5pV3nn61un9zzHuXB1lMhHBHzgOuB1wM7gG9HxMbM3DpTv2OqLodu/pdUOpJuu7Om+j+wyYuBzivmmQjoOt7jWN+nTjNV42z/o9hTIQycC2zLzB8CRMTtwCXAjIVwp5+N7+CpF78IgGd+vJt5/7ifp178ommXD7ffsW6r+/393fX97n1j22k2W8vtd1t0/tnq3O8vPvs1XnzyCAA/fnQrL1m68tBxnXdtTO778yfH+Pd/+LpfeoikGzP9HsAxv0+dZqrGzv9en/2L99TajRS9dAdARLwNuDgz312tvwP47cx8b9s+a4G11erLgUdmvdBj81Jgb+kiChjEzz2InxkG83NP95n3ZubF3bxBr10JH1FmbgA2lK7jaEXElsxcU7qO2TaIn3sQPzMM5ueeic/ca9Mb7QSWtK2PVm2S1Jd6LYS/DayIiNMj4vnAZcDGwjVJUm16qjsiMw9ExHuBr9C6Re3GzHy4cFkzZc51ocyQQfzcg/iZYTA/93F/5p76Yk6SBk2vdUdI0kAxhCWpIEO4BhGxJCLui4itEfFwRLyvaj85Iu6JiB9UP08qXetMi4h5EfG9iPhStX56RNwfEdsi4nPVF659JSIWRsSdEfH3EdGIiPP7/VxHxL+q/mw/FBG3RcQL+/FcR8SNEbEnIh5qa5vy3EbLp6rP/2BEnNPN7zCE63EA+EBmngmcB6yLiDOBq4B7M3MFcG+13m/eB7Q/O/px4LrMXA78BLiiSFX1+iRwd2a+AngVrc/ft+c6IhYDfwqsycxX0voS/TL681zfBHQ+dDHduX0jsKJ6rQXWd/UbMtNXzS/gi7TGw3gEWFS1LQIeKV3bDH/O0eoP5YXAl4Cg9TTR86rt5wNfKV3nDH/mE4FHqb7kbmvv23MNLAYeB06mdYfVl4A39Ou5BpYBDx3p3AL/BXj7VPsd7uWVcM0iYhlwNnA/MJKZu6pNu4GRQmXV5a+Afws8W62fAjyVmQeq9R20/gL3k9OBceBvqm6Yz0TECfTxuc7MncAngO3ALuBp4Dv0/7meNN25nfzHaVJX/w0M4RpFxHzg88D7M/On7duy9U9l39wfGBFvBvZk5ndK1zLLngecA6zPzLOBZ+joeujDc30SrYG1TgdeBpzAr/4v+0CYiXNrCNckIoZoBfCtmXlX1TwWEYuq7YuAPaXqq8FrgH8WEY8Bt9PqkvgksDAiJh8K6sfH0HcAOzLz/mr9Tlqh3M/n+nXAo5k5npkTwF20zn+/n+tJ053bYxp2wRCuQUQEcAPQyMxr2zZtBC6vli+n1VfcFzLz6swczcxltL6k+Xpm/gFwH/C2are++swAmbkbeDwiXl41XURr6NW+Pde0uiHOi4gXV3/WJz9zX5/rNtOd243AO6u7JM4Dnm7rtpiWT8zVICJeC/wf4Ps81z/6IVr9wncAS4EfAZdm5pNFiqxRRFwA/JvMfHNE/AatK+OTge8Bf5iZ+0vWN9MiYjXwGeD5wA+Bd9G6wOnbcx0RHwF+n9adQN8D3k2r/7OvznVE3AZcQGvIyjHgw8D/YIpzW/2D9Ne0umZ+DrwrM7cc8XcYwpJUjt0RklSQISxJBRnCklSQISxJBRnCklSQIayBEBHL2kfCOspjXxYRd850TRL02PRGUi/KzCd47iEEaUZ5JaxB8ryIuLUa8/fO6omvxyLimoh4ICK2RMQ5EfGViPiHiPgTOL6raOlIDGENkpcD/ykzVwI/Bd5TtW/PzNW0nnK8idZV73nAR0oUqcFid4QGyeOZ+X+r5c/SGpgcWs/8Q+sx8/mZuQ/YFxH7I2LhbBepweKVsAZJ5zP6k+uT4xs827Y8ue6FimplCGuQLI2I86vlfwF8s2QxEhjCGiyP0JrvrwGcRLdzgEk1chQ1SSrIK2FJKsgQlqSCDGFJKsgQlqSCDGFJKsgQlqSCDGFJKuj/Ay/pZB8ZtCX+AAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 360x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wAX8xBJNVfar"
      },
      "source": [
        "Dai precedenti grafici per le distribuzioni delle feature `avg_glucose_level` e `bmi` osserviamo che:\n",
        "* Il livello medio di glucosio è molto concentrato nell'intervallo che va circa da 60 a 100\n",
        "* Il bmi è molto concentrato nell'intevallo [20;40]"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 606
        },
        "id": "nZlvLNIiiWvC",
        "outputId": "c124466a-7714-45c2-a831-b3960e52d50e"
      },
      "source": [
        "plt.figure(figsize=(15,10))\n",
        "\n",
        "plt.subplot(1,2,1)\n",
        "sb.histplot(x=\"gender\", data=stroke_dataset);\n",
        "\n",
        "plt.subplot(1,2,2)\n",
        "sb.histplot(x=\"age\", data=stroke_dataset);"
      ],
      "execution_count": 106,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAA4EAAAJNCAYAAACP93C3AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzde7hlZ10n+O8vqXARkICUeUKl0mE0ioEZApYhBKcngkDI2BMYkYRxIGLsxGeCYDftCHb3oK3pxh4RwVY60URCD5JELhI1DcZwGyZyqWAMJIGmhECqCEk1hIvShKeS3/yxV8GhqMupqnPOPue8n8/z7Oes9a7L/u19dtV7vnut9a7q7gAAADCGI+ZdAAAAACtHCAQAABiIEAgAADAQIRAAAGAgQiAAAMBAhEAAAICBbJh3AcvhEY94RJ9wwgnzLgOAFXDDDTf81+7eOO861gp9JMAY9tc/rssQeMIJJ2Tr1q3zLgOAFVBVn5l3DWuJPhJgDPvrH50OCgAAMBAhEAAAYCDLFgKr6gFV9aGq+tuqurmqfm1qf1RVfbCqtlXVlVV1v6n9/tP8tmn5CQv29fKp/RNV9YzlqhkAAGC9W84jgfckeUp3Py7JyUnOqKpTk/xmkld39/cnuTvJedP65yW5e2p/9bRequqkJOckeUySM5L8flUduYx1AwAArFvLFgJ75u+n2aOmRyd5SpI3T+2XJ3nWNH3WNJ9p+VOrqqb2K7r7nu7+dJJtSU5ZrroBAADWs2W9JrCqjqyqG5PcleTaJH+X5EvdvWtaZXuSTdP0piS3J8m0/MtJvmdh+162AQAA4CAsawjs7nu7++Qkx2V29O7Ry/VcVXV+VW2tqq07d+5crqcBAABY01ZkdNDu/lKSdyd5UpKjq2r3/QmPS7Jjmt6RZHOSTMsfmuQLC9v3ss3C57iku7d095aNG90zGAAAYG+Wc3TQjVV19DT9wCRPS3JrZmHwOdNq5yZ5+zR99TSfafm7urun9nOm0UMfleTEJB9arroBAADWsw0HXuWQHZvk8mkkzyOSXNXdf15VtyS5oqp+I8nfJLl0Wv/SJP+pqrYl+WJmI4Kmu2+uqquS3JJkV5ILu/veZawbAABg3Vq2ENjdNyV5/F7aP5W9jO7Z3V9P8lP72NdFSS5a6hoBAABGsyLXBAIAALA6CIEAAAADEQIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwECEQAABgIEIgAADAQIRAAACAgQiBAAAAAxECAQAABiIEAgAADEQIBAAAGMiGeRcAK2HT5uPzue23z7sMlsAjj9ucHbd/dt5lALBClqIP13fAtxMCGcLntt+esy++ft5lsASuvOC0eZcAwApaij5c3wHfzumgALBKVNUDqupDVfW3VXVzVf3a1P76qvp0Vd04PU6e2quqXltV26rqpqp6wnxfAQBrgSOBALB63JPkKd3991V1VJL3V9V/npb9Une/eY/1n5nkxOnxxCSvm34CMCdr4RRmIRAAVonu7iR/P80eNT16P5ucleQN03YfqKqjq+rY7r5jmUsFYB/WwinMTgcFgFWkqo6sqhuT3JXk2u7+4LTooumUz1dX1f2ntk1JFn7dvH1qA4B9EgIBYBXp7nu7++QkxyU5paoem+TlSR6d5EeSPDzJLx/MPqvq/KraWlVbd+7cueQ1A7C2CIEAsAp195eSvDvJGd19R8/ck+SPkpwyrbYjyeYFmx03te25r0u6e0t3b9m4ceNylw7AKicEAsAqUVUbq+roafqBSZ6W5ONVdezUVkmeleRj0yZXJ3nBNEroqUm+7HpAAA7EwDAAsHocm+Tyqjoysy9qr+ruP6+qd1XVxiSV5MYkPz+tf02SM5NsS/K1JC+cQ80ArDFCIACsEt19U5LH76X9KftYv5NcuNx1AbC+OB0UAABgIEIgAADAQIRAAACAgQiBAAAAAxECAQAABiIEAgAADEQIBAAAGIgQCAAAMBAhEAAAYCBCIAAAwECEQAAAgIEIgQAAAAMRAgEAAAYiBAIAAAxECAQAABiIEAgAAGvIps3Hp6oO67Fp8/HzfhnM0YZ5FwAAACze57bfnrMvvv6w9nHlBactUTWsRY4EAgAADEQIBAAAGIgQCAAAMBAhEAAAYCBCIAAArJClGNkTDpfRQQEAYIUY2ZPVwJFAAACAgQiBAAAAAxECAQAABiIEAgAADEQIBAAAGIgQCAAAMBAhEAAAYCBCIAAAwECEQAAAgIEIgQAAAAMRAgEAAAYiBAIAAAxECAQAABiIEAgAADAQIRAAAGAgQiAAAMBAhEAAAICBCIEAAAADEQIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwECEQAABgIEIgAADAQIRAAACAgQiBAAAAAxECAQAABiIEAgAADEQIBAAAGIgQCAAAMBAhEAAAYCBCIAAAwECEQAAAgIEsWwisqs1V9e6quqWqbq6ql0ztv1pVO6rqxulx5oJtXl5V26rqE1X1jAXtZ0xt26rqZctVMwAAwHq3YRn3vSvJS7v7I1X1kCQ3VNW107JXd/dvLVy5qk5Kck6SxyR5ZJK/qqofmBb/XpKnJdme5MNVdXV337KMtQMAAKxLyxYCu/uOJHdM01+tqluTbNrPJmcluaK770ny6araluSUadm27v5UklTVFdO6QiAAAMBBWpFrAqvqhCSPT/LBqelFVXVTVV1WVQ+b2jYluX3BZtuntn21AwAAcJCWPQRW1YOTvCXJL3b3V5K8Lsn3JTk5syOFr1qi5zm/qrZW1dadO3cuxS4BAADWnWUNgVV1VGYB8I3d/dYk6e47u/ve7r4vyR/kW6d87kiyecHmx01t+2r/Nt19SXdv6e4tGzduXPoXAwAAsA4s5+igleTSJLd2928vaD92wWrPTvKxafrqJOdU1f2r6lFJTkzyoSQfTnJiVT2qqu6X2eAxVy9X3QAAAOvZco4O+uQkz0/y0aq6cWr7lSTPq6qTk3SS25JckCTdfXNVXZXZgC+7klzY3fcmSVW9KMk7kxyZ5LLuvnkZ6wYAAFi3lnN00Pcnqb0sumY/21yU5KK9tF+zv+0AYD2oqgckeV+S+2fWR7+5u18xnSFzRZLvSXJDkud39zeq6v5J3pDkh5N8IcnZ3X3bXIoHYM1YkdFBAYBFuSfJU7r7cZkNoHZGVZ2a5Dczu8fu9ye5O8l50/rnJbl7an/1tB4A7JcQCACrRM/8/TR71PToJE9J8uap/fIkz5qmz5rmMy1/6nRNPgDskxAIAKtIVR05XUt/V5Jrk/xdki91965plYX3y/3mvXSn5V/O7JRRANgnIRAAVpHpNkonZ3ZLpFOSPPpw9+leugAsJAQCwCrU3V9K8u4kT0pydFXtHsxt4f1yv3kv3Wn5QzMbIGbPfbmXLgDfJAQCwCpRVRur6uhp+oFJnpbk1szC4HOm1c5N8vZp+uppPtPyd3V3r1zFAKxFy3mfQADg4Byb5PKqOjKzL2qv6u4/r6pbklxRVb+R5G+SXDqtf2mS/1RV25J8Mck58ygagLVFCASAVaK7b0ry+L20fyqz6wP3bP96kp9agdIAWEecDgoAADAQIRAAAGAgQiAAAMBAhEAAAICBCIEAAAADEQIBAAAGIgQCAAAMRAgEAGB9O2JDquqwH5s2Hz/vVwJLws3iAQBY3+7blbMvvv6wd3PlBactQTEwf44EAgAADEQIBAAAGIgQCAAAMBAhEAAAYCBCIAAAwECEQAAAgIEIgQAAAAMRAgEAAAYiBAIAAAxECAQAABiIEAgAADAQIRAAAGAgQiAAAMBAhEAAAICBCIEAAAADEQIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwECEQAABgIEIgAADAQIRAAACAgQiBAAAAAxECAQBgNEdsSFUd9mPT5uPn/Uo4BBvmXQAAALDC7tuVsy++/rB3c+UFpy1BMaw0RwIBAAAGIgQCAAAMxOmgAACwGNN1dLDWCYEAALAYS3AdnWvoWA2cDgoAADAQIRAAAGAgQiAAAMBAhEAAAICBCIEAAAADEQIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwECEQAABgIEIgAADAQIRAAACAgQiBAAAAAxECAQAABiIEAgAADEQIBAAAGIgQCAAAMBAhEAAAYCBCIAAAwECEQAAAgIEIgQAAAAMRAgEAAAYiBAIAAAxECAQAABiIEAgAAByaIzakqg7rsWnz8fN+FcPZMO8CAACANeq+XTn74usPaxdXXnDaEhXDYjkSCACrRFVtrqp3V9UtVXVzVb1kav/VqtpRVTdOjzMXbPPyqtpWVZ+oqmfMr3oA1gpHAgFg9diV5KXd/ZGqekiSG6rq2mnZq7v7txauXFUnJTknyWOSPDLJX1XVD3T3vStaNQBriiOBALBKdPcd3f2RafqrSW5Nsmk/m5yV5Iruvqe7P51kW5JTlr9SANYyIRAAVqGqOiHJ45N8cGp6UVXdVFWXVdXDprZNSW5fsNn27D80AoAQCACrTVU9OMlbkvxid38lyeuSfF+Sk5PckeRVB7m/86tqa1Vt3blz55LXC8DaIgQCwCpSVUdlFgDf2N1vTZLuvrO77+3u+5L8Qb51yueOJJsXbH7c1PZtuvuS7t7S3Vs2bty4vC8AgFVPCASAVaKqKsmlSW7t7t9e0H7sgtWeneRj0/TVSc6pqvtX1aOSnJjkQytVLwBrk9FBAWD1eHKS5yf5aFXdOLX9SpLnVdXJSTrJbUkuSJLuvrmqrkpyS2Yji15oZFAADmTZQmBVbU7yhiTHZNZpXdLdr6mqhye5MskJmXVkz+3uu6dvP1+T5MwkX0vyM7tHSKuqc5P8q2nXv9Hdly9X3QAwL939/iS1l0XX7Gebi5JctGxFAbDuLOfpoLvvdXRSklOTXDjdz+hlSa7r7hOTXDfNJ8kzMzuN5cQk52d2EXym0PiKJE/M7BqIVywYFQ0AWGGbNh+fqjrsx6bNx8/7pQAMadmOBHb3HZmNYJbu/mpV7b7X0VlJTp9WuzzJe5L88tT+hu7uJB+oqqOnayBOT3Jtd38xSaab5p6R5E3LVftumzYfn89tv/3AKwLAQD63/facffH1h72fKy84bQmqAeBgrcg1gXvc6+iYKSAmyeczO1002fe9juZ2D6Sl6uSYP39oAMDiLdUX4Y88bnN23P7ZJagIWErLHgL3vNfR7NK/me7uquolep7zMzuNNMcf7/QSAIBD5WgvrG/LeouIvd3rKMmdu4e6nn7eNbXv615H7oEEAACwRJYtBO7rXkeZ3dPo3Gn63CRvX9D+gpo5NcmXp9NG35nk6VX1sGlAmKdPbQAAAByk5TwddF/3Onplkquq6rwkn0ny3GnZNZndHmJbZreIeGGSdPcXq+rXk3x4Wu/f7B4kBgAAgIOznKOD7uteR0ny1L2s30ku3Me+Lkty2dJVBwAAMKZlvSYQAACA1UUIBAAAGIgQCAAAMBAhEAAAYCBCIAAAwECEQAAAgIEIgQAAAAMRAgEAAAYiBAIAAAxECAQAABiIEAgAADAQIRAAAGAgQiAAAMBAhEAAAICBCIEAAAADEQIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwECEQAABgIEIgAADAQIRAAACAgQiBAAAAAxECAQAABiIEAgCsI5s2H5+qOqwHsL5tmHcBAAAsnc9tvz1nX3z9Ye3jygtOW6JqgNXIkUAAAICBCIEAAAADEQIBAA7TUlyHV1XZtPn4eb+UpXXEBtcnwirkmkAAgMO0FNfhJevwWrz7drk+EVYhRwIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwECEQAABgIEIgAADAQIRAAACAgQiBAAAAAxECAQAABiIEAgAADEQIBAAAGIgQCAAAMBAhEAAAYCBCIAAAwECEQAAAgIEIgQAAAAMRAgEAAAayYd4FAAAwOWJDqmreVQDrnBAIALBa3LcrZ198/WHt4soLTluiYoD1yumgAAAAAxECAQAABiIEAgAADEQIBAAAGIgQCAAAMBAhEAAY2qbNx6eqDusBsJa4RQQAMLTPbb/dbRmAoTgSCACrRFVtrqp3V9UtVXVzVb1kan94VV1bVZ+cfj5saq+qem1Vbauqm6rqCfN9BQCsBUIgAKweu5K8tLtPSnJqkgur6qQkL0tyXXefmOS6aT5JnpnkxOlxfpLXrXzJAKw1QiAArBLdfUd3f2Sa/mqSW5NsSnJWksun1S5P8qxp+qwkb+iZDyQ5uqqOXeGyAVhjhEAAWIWq6oQkj0/ywSTHdPcd06LPJzlmmt6U5PYFm22f2gBgn4RAAFhlqurBSd6S5Be7+ysLl3V3J+mD3N/5VbW1qrbu3LlzCSsFYC0SAgFgFamqozILgG/s7rdOzXfuPs1z+nnX1L4jyeYFmx83tX2b7r6ku7d095aNGzcuX/EArAlCIACsEjW74dylSW7t7t9esOjqJOdO0+cmefuC9hdMo4SemuTLC04bBYC9cp9AAFg9npzk+Uk+WlU3Tm2/kuSVSa6qqvOSfCbJc6dl1yQ5M8m2JF9L8sKVLRdgCRyxIbPvwA7dI4/bnB23f3aJClr/hEAAWCW6+/1J9vWX0FP3sn4nuXBZiwJYbvftytkXX39Yu7jygtOWqJgxOB0UAABgIEIgAADAQIRAAACAgSwqBFbVkxfTBgDoNwFY3RZ7JPB3F9kGAOg3AVjF9js6aFU9KclpSTZW1T9fsOi7kxy5nIUBwFqj3wRgLTjQLSLul+TB03oPWdD+lSTPWa6iAGCN0m8CsOrtNwR293uTvLeqXt/dn1mhmgBgTdJvArAWLPZm8fevqkuSnLBwm+5+ynIUBQBrnH4TgFVrsSHwT5L8xyR/mOTe5SsHANYF/SYAq9ZiQ+Cu7n7dslYCAOuHfhOAVWuxt4j4s6r6P6rq2Kp6+O7HslYGAGuXfhOAVWuxRwLPnX7+0oK2TvLfLW05ALAu6DcBWLUWFQK7+1HLXQgArBf6TQBWs0WFwKp6wd7au/sNS1sOAKx9+k0AVrPFng76IwumH5DkqUk+kkRnBgDfSb8JwKq12NNBf2HhfFUdneSK/W1TVZcl+Ykkd3X3Y6e2X03yT5PsnFb7le6+Zlr28iTnZTaU9ou7+51T+xlJXpPkyCR/2N2vXNQrA4A5OZR+EwBWymJHB93TPyQ50PUOr09yxl7aX93dJ0+P3QHwpCTnJHnMtM3vV9WRVXVkkt9L8swkJyV53rQuAKwli+k3AWBFLPaawD/LbFSzZHZE7oeSXLW/bbr7fVV1wiLrOCvJFd19T5JPV9W2JKdMy7Z196emOq6Y1r1lkfsFgBV3KP0mAKyUxV4T+FsLpncl+Ux3bz/E53zRdMH81iQv7e67k2xK8oEF62yf2pLk9j3an3iIzwsAK2Up+00AWFKLOh20u9+b5ONJHpLkYUm+cYjP97ok35fk5CR3JHnVIe7nO1TV+VW1taq27ty588AbAMAyWcJ+EwCW3KJCYFU9N8mHkvxUkucm+WBVPedgn6y77+zue7v7viR/kG+d8rkjyeYFqx43te2rfW/7vqS7t3T3lo0bNx5saQCwZJaq3wSA5bDY00H/ZZIf6e67kqSqNib5qyRvPpgnq6pju/uOafbZST42TV+d5I+r6reTPDLJiZl1npXkxKp6VGbh75wk/9vBPCcAzMGS9JsAsBwWGwKP2N2RTb6QAxxFrKo3JTk9ySOqanuSVyQ5vapOzuxi+duSXJAk3X1zVV2V2YAvu5Jc2N33Tvt5UZJ3ZnZh/WXdffMiawaAeTnofhMAVspiQ+A7quqdSd40zZ+d5Jr9bdDdz9tL86X7Wf+iJBftpf2aAz0XAKwyB91vAsBK2W8IrKrvT3JMd/9SVf2vSX50WvTXSd643MUBwFqi3wRgLTjQkcDfSfLyJOnutyZ5a5JU1X8/Lfsny1odAKwt+k0AVr0DXZ9wTHd/dM/Gqe2EZakIANYu/SbAPByxIVV12I9Nm4+f9ytZEQc6Enj0fpY9cCkLAYB1QL8JMA/37crZF19/2Lu58oLTlqCY1e9ARwK3VtU/3bOxqn4uyQ3LUxIArFn6TQBWvQMdCfzFJG+rqp/OtzqvLUnul9l9/gCAb9FvArDq7TcEdvedSU6rqh9L8tip+S+6+13LXhkArDH6TQDWgkXdJ7C7353k3ctcCwCsC/rNRZoGcjgcjzxuc3bc/tklKghgDIu9WTwAwNJagoEcRhnEAWApHWhgGAAAANYRIRAAAGAgQiAAAMBAhEAAAICBCIEAAAADEQIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwECEQAABgIEIgAADAQIRAAACAgWyYdwEAAACrwhEbUlXzrmLZCYEAAABJct+unH3x9Ye1iysvOG2Jilk+TgcFAAAYiBAIAAAwECEQAABgIEIgAADAQIRAAACAgQiBAAAAAxECAQAABiIEAgAADEQIBAAAGIgQCAAAMBAhEAAAYCBCIAAAwECEQAAAgIEIgQAAAAMRAgEAAAYiBAIAAAxECASAVaKqLququ6rqYwvafrWqdlTVjdPjzAXLXl5V26rqE1X1jPlUDcBaIwQCwOrx+iRn7KX91d198vS4Jkmq6qQk5yR5zLTN71fVkStWKQBrlhAIAKtEd78vyRcXufpZSa7o7nu6+9NJtiU5ZdmKW62O2JCqOqwHwGg2zLsAAOCAXlRVL0iyNclLu/vuJJuSfGDBOtuntrHctytnX3z9Ye3iygtOW6JiANYGRwIBYHV7XZLvS3JykjuSvOpgd1BV51fV1qraunPnzqWuD4A1RggEgFWsu+/s7nu7+74kf5BvnfK5I8nmBaseN7XtbR+XdPeW7t6ycePG5S0YgFVPCASAVayqjl0w++wku0cOvTrJOVV1/6p6VJITk3xopesDYO1xTSAArBJV9aYkpyd5RFVtT/KKJKdX1clJOsltSS5Iku6+uaquSnJLkl1JLuzue+dRNwBrixAIAKtEdz9vL82X7mf9i5JctHwVAbAeOR0UAABgIEIgAADAQIRAAACAgQiBAAAAAxECAQAABiIEAgAADEQIBAAAGIgQCAAAMBAhEAAAYCBCIAAAwECEQAAAgIEIgQAAAAMRAgEAAAYiBAIAAAxECAQAABiIEAgAADAQIRAAAGAgQiAAAMBAhEAAAICBCIEAAAADEQIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwECEQAABgIEIgAADAQIRAAACAgQiBAAAAAxECAQAABiIEAgAADEQIBAAAGIgQCAAAMJBlC4FVdVlV3VVVH1vQ9vCquraqPjn9fNjUXlX12qraVlU3VdUTFmxz7rT+J6vq3OWqFwAAYATLeSTw9UnO2KPtZUmu6+4Tk1w3zSfJM5OcOD3OT/K6ZBYak7wiyROTnJLkFbuDIwAAAAdv2UJgd78vyRf3aD4ryeXT9OVJnrWg/Q0984EkR1fVsUmekeTa7v5id9+d5Np8Z7AEAABgkVb6msBjuvuOafrzSY6ZpjcluX3Betuntn21AwAAcAjmNjBMd3eSXqr9VdX5VbW1qrbu3LlzqXYLAACwrqx0CLxzOs0z08+7pvYdSTYvWO+4qW1f7d+huy/p7i3dvWXjxo1LXjgAAMB6sNIh8Ooku0f4PDfJ2xe0v2AaJfTUJF+eTht9Z5KnV9XDpgFhnj61AQAAcAg2LNeOq+pNSU5P8oiq2p7ZKJ+vTHJVVZ2X5DNJnjutfk2SM5NsS/K1JC9Mku7+YlX9epIPT+v9m+7ec7AZAAAAFmnZQmB3P28fi566l3U7yYX72M9lSS5bwtIAAACGNbeBYQAAAFh5QiAAAMBAhEAAAICBCIEAAAADEQIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwECEQAABgIEIgAADAQIRAAACAgQiBAAAAAxECAQAABiIEAgAADEQIBAAAGIgQCAAAMBAhEAAAYCBCIAAAwECEQAAAgIEIgQAAAAMRAgEAAAYiBAIAAAxECAQAABiIEAgAADAQIRAAVomquqyq7qqqjy1oe3hVXVtVn5x+Pmxqr6p6bVVtq6qbquoJ86scgLVECASA1eP1Sc7Yo+1lSa7r7hOTXDfNJ8kzk5w4Pc5P8roVqhGANU4IBIBVorvfl+SLezSfleTyafryJM9a0P6GnvlAkqOr6tiVqRSAtUwIBIDV7ZjuvmOa/nySY6bpTUluX7De9qkNAPZLCASANaK7O0kf7HZVdX5Vba2qrTt37lyGygBYS4RAAFjd7tx9muf0866pfUeSzQvWO25q+w7dfUl3b+nuLRs3blzWYgFY/YRAAFjdrk5y7jR9bpK3L2h/wTRK6KlJvrzgtFEA2KcN8y4AAJipqjclOT3JI6pqe5JXJHllkquq6rwkn0ny3Gn1a5KcmWRbkq8leeGKFwzAmiQEAsAq0d3P28eip+5l3U5y4fJWBMB65HRQAACAgQiBAAAAAxECAQAABiIEAgAADEQIBAAAGIgQCAAAMBAhEAAAYCBCIAAAwECEQAAAgIEIgQAAAAMRAgEAAAYiBAIAAAxECAQAABiIEAgAADAQIRAAAGAgQiAAAMBAhEAAAICBCIEAAAADEQIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwECEQAABgIEIgAADAQIRAAACAgQiBAAAAAxECAQAABiIEAgAADEQIBAAAGIgQCAAAMBAhEAAAYCBCIAAAwECEQAAAgIEIgQAAAAMRAgEAAAYiBAIAAAxECAQAABiIEAgAADAQIRAAAGAgQiAAAMBAhEAAAICBCIEAAAADEQIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwECEQAABgIHMJgVV1W1V9tKpurKqtU9vDq+raqvrk9PNhU3tV1WuraltV3VRVT5hHzQAAAOvBPI8E/lh3n9zdW6b5lyW5rrtPTHLdNJ8kz0xy4vQ4P8nrVrxSAACAdWI1nQ56VpLLp+nLkzxrQfsbeuYDSY6uqmPnUSAAAMBaN68Q2En+sqpuqKrzp7ZjuvuOafrzSY6ZpjcluX3BttunNgAAAA7Shjk97492946q+t4k11bVxxcu7O6uqj6YHU5h8vwkOf7445euUgAAgHVkLkcCu3vH9POuJG9LckqSO3ef5jn9vGtafUeSzeTLxoIAAA8tSURBVAs2P25q23Ofl3T3lu7esnHjxuUsHwAAYM1a8RBYVQ+qqofsnk7y9CQfS3J1knOn1c5N8vZp+uokL5hGCT01yZcXnDYKAADAQZjH6aDHJHlbVe1+/j/u7ndU1YeTXFVV5yX5TJLnTutfk+TMJNuSfC3JC1e+ZAAAgPVhxUNgd38qyeP20v6FJE/dS3snuXAFSgMAAFj3VtMtIgAAAFhmQiAAAMBAhEAAAICBCIEAAAADEQIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwkBW/WTwAcPCq6rYkX01yb5Jd3b2lqh6e5MokJyS5Lclzu/vuedUIwNrgSCAArB0/1t0nd/eWaf5lSa7r7hOTXDfNA8B+CYEAsHadleTyafryJM+aYy0ArBFCIACsDZ3kL6vqhqo6f2o7prvvmKY/n+SY+ZQGwFrimkAAWBt+tLt3VNX3Jrm2qj6+cGF3d1X13jacQuP5SXL88ccvf6UArGqOBALAGtDdO6afdyV5W5JTktxZVccmyfTzrn1se0l3b+nuLRs3blypkgFYpYRAAFjlqupBVfWQ3dNJnp7kY0muTnLutNq5Sd4+nwoBWEucDgoAq98xSd5WVcms7/7j7n5HVX04yVVVdV6SzyR57hxrBGCNEAIBYJXr7k8ledxe2r+Q5KkrXxEAa5nTQQEAAAYiBAIAAAxECAQAABiIEAgAADAQIRAAAGAgQiAAAMBAhEAAAICBCIEAAAADEQIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwECEQAABgIEIgAADAQIRAAACAgQiBAAAAAxECAQAABiIEAgAADEQIBAAAGIgQCAAAMBAhEAAAYCBCIAAAwECEQAAAgIEIgQAAAAMRAgEAAAYiBAIAAAxECAQAABiIEAgAADAQIRAAAGAgQiAAAMBAhEAAAICBCIEAAAADEQIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwECEQAABgIEIgAADAQIRAAACAgQiBAAAAAxECAQAABiIEAgAADEQIBAAAGIgQCAAAMBAhEAAAYCBCIAAAwECEQAAAgIEIgQAAAAMRAgEAAAYiBAIAAAxECAQAABiIEAgAADAQIRAAAGAgQiAAAMBAhEAAAICBCIEAAAADEQIBAAAGIgQCAAAMRAgEAAAYiBAIAAAwECEQAABgIEIgAADAQIRAAACAgQiBAAAAA1kzIbCqzqiqT1TVtqp62bzrAYDVQP8IwMFaEyGwqo5M8ntJnpnkpCTPq6qT5lsVAMyX/hGAQ7EmQmCSU5Js6+5Pdfc3klyR5Kw51wQA86Z/BOCgrZUQuCnJ7Qvmt09tADAy/SMAB626e941HFBVPSfJGd39c9P885M8sbtftGCd85OcP83+YJJPrHiha9MjkvzXeRfBkvH7XD/8LhfvH3X3xnkXMQ+L6R+n9qXuI30+l5b3c+l5T5eW93PprcR7us/+ccMyP/FS2ZFk84L546a2b+ruS5JcspJFrQdVtbW7t8y7DpaG3+f64XfJIh2wf0yWvo/0+Vxa3s+l5z1dWt7PpTfv93StnA764SQnVtWjqup+Sc5JcvWcawKAedM/AnDQ1sSRwO7eVVUvSvLOJEcmuay7b55zWQAwV/pHAA7FmgiBSdLd1yS5Zt51rENOoV1f/D7XD79LFmVO/aPP59Lyfi497+nS8n4uvbm+p2tiYBgAAACWxlq5JhAAAIAlIASuU1XVVfX/LJjfUFU7q+rPD7Dd6Qdah+VRVfdW1Y0LHics43PdVlWPWK798+2q6riqentVfbKq/q6qXlNV96uqk6vqzAXr/WpV/Yt51gr7U1VnVNUnqmpbVb1s3vWsRVW1uareXVW3VNXNVfWSqf3hVXXt9P/EtVX1sHnXupZU1ZFV9Te7/4aZBkv64PRZvXIaOIlFqqqjq+rNVfXxqrq1qp7kM3roquqfTf/eP1ZVb6qqB8z7MyoErl//kOSxVfXAaf5p2cuw4awq/627T17wuG3eBXH4qqqSvDXJn3b3iUl+IMmDk1yU5OQkZ+5n84N9riOXal+wp+nz9XtJnpnkpCTPq6qT5lvVmrQryUu7+6Qkpya5cHofX5bkuun/ieumeRbvJUluXTD/m0le3d3fn+TuJOfNpaq16zVJ3tHdj07yuMzeW5/RQ1BVm5K8OMmW7n5sZoN4nZM5f0aFwPXtmiT/8zT9vCRv2r2gqk6pqr+evjW7vqp+cM+Nq+pBVXVZVX1oWu+sFaqbSVX9cFW9t6puqKp3VtWxU/t7qurVVbV1+obuR6rqrdO3c7+xYPs/nba9ebpZ9N6e43+ffsc3VtXFgsSSe0qSr3f3HyVJd9+b5J8l+bkk/z7J2dN7f/a0/knT7/dTVfXi3TvZ1++pqv6+ql5VVX+b5Ekr+soYzSlJtnX3p7r7G0muSKJfOEjdfUd3f2Sa/mpmf1xvyuy9vHxa7fIkz5pPhWtPVR2X2d87fzjNV2b/9755WsX7eRCq6qFJ/nGSS5Oku7/R3V+Kz+jh2JDkgVW1Icl3Jbkjc/6MCoHr2xVJzqmqByT5H5J8cMGyjyf5H7v78Un+ryT/di/b/8sk7+ruU5L8WJL/u6oetMw1j+yBC04FfVtVHZXkd5M8p7t/OMllmR092u0b001G/2OStye5MMljk/xMVX3PtM7PTttuSfLiBe1Jkqr6oSRnJ3lyd5+c5N4kP72Mr3FEj0lyw8KG7v5KktuS/EaSK6cjv1dOix+d5BmZ/cH9iqo66gC/pwcl+WB3P66737/sr4aRbUpy+4L57VMbh2g67f/xmfXPx3T3HdOizyc5Zk5lrUW/k+T/THLfNP89Sb7U3bumeZ/Vg/OoJDuT/NF0EOAPp7//fEYPQXfvSPJbST6bWfj7cmZ/F8z1M7pmbhHBwevum6YO5nn5zuHDH5rk8qo6MUknOWovu3h6kv9lwTVKD0hyfL79dAuWzn+b/sBPklTVYzMLddfOvtTMkZn957Hb7htCfzTJzbv/Y66qTyXZnOQLmQW/Z0/rbU5y4tS+21OT/HCSD0/P8cAkdy3ty+Ig/UV335Pknqq6K7NOdn+/p3uTvGUehQKHrqoenNm/3V/s7q9M/7aTJN3dVWX49kWoqp9Icld331BVp8+7nnViQ5InJPmF7v5gVb0me5z66TO6eNO1k2dlFq6/lORPkpwx16IiBI7g6sy+fTg9s2/Gdvv1JO/u7mdPQfE9e9m2kvxkd39ieUtkHyqzcLevU/zumX7et2B69/yGqTP88SRP6u6vVdV7Mgvyez7H5d398iWrmj3dkuQ5Cxuq6rsz+0Jl117WX/i7vDez/6f393v6+nSKKSy3HZl9mbTbcXGt+SGZzvR4S5I3dvdbp+Y7q+rY7r5jOvXfF3KL8+TMvrA+M7M+7rszu57t6KraMB1p8Vk9ONuTbO/u3WeQvTmzEOgzemh+PMmnu3tnklTVWzP73M71M+p00PXvsiS/1t0f3aP9ofnWh+1n9rHtO5P8wnRufarq8ctSIfvyiSQbq+pJyeyPhqp6zEFs/9Akd08B8NGZDUCwp+uSPKeqvnd6jodX1T863ML5Ntcl+a6qekHyzcE1XpXk9UnuTPKQRe7D74l5+3CSE6cR7e6X2cAGVx9gG/Yw9amXJrm1u397waKrk5w7TZ+b2Wn+HEB3v7y7j+vuEzL7TL6ru386ybvzrS/gvJ8Hobs/n+T2BeNFPDWzLzR9Rg/NZ5OcWlXfNf373/1+zvUzKgSuc929vbtfu5dF/z7Jv6uqv8m+jwj/emanid5UVTdP86yQaeCF5yT5zWnQjxuTnHYQu3hHZkcEb03yyiQf2Mtz3JLkXyX5y6q6Kcm1SY493Nr5lu7uJM9O8lNV9ckk/yXJ15P8SmYdwEl7DAyzt334PTF307fVL8rsC8Jbk1zV3TfPt6o16clJnp/kKQuuAz8zs/+nnzb9P/Hj0zyH7peT/POq2pbZmVCXzrmeteYXkrxx6nNOzmzsCJ/RQzAdUX1zko9kdgnPEUkuyZw/ozX7+wQAAIAROBIIAAAwECEQAABgIEIgAADAQIRAAACAgQiBAAAAAxECYR2rqtdX1XMOvCYAAKMQAoFvqqp93TMSAIB1QgiEVaKq/nVVfaKq3l9Vb6qqf1FV31dV76iqG6rq/62qR0/rvr6qXltV11fVp3Yf7auZ/zDt56+SfO+C/f9wVb132tc7q+rYqf09VfU7VbU1yUvm8doB4HBV1Z9OfdzNVXX+1HZeVf2XqvpQVf1BVf2HqX1jVb2lqj48PZ483+phZfnWH1aBqvqRJD+Z5HFJjkrykSQ3JLkkyc939yer6olJfj/JU6bNjk3yo0keneTqJG9O8uwkP5jkpCTHJLklyWVVdVSS301yVnfvrKqzk1yU5Genfd2vu7cs+wsFgOXzs939xap6YJIPV9VfJPnXSZ6Q5KtJ3pXkb6d1X5Pk1d39/qo6Psk7k/zQPIqGeRACYXV4cpK3d/fXk3y9qv4syQOSnJbkT6pq93r3X7DNn3b3fUluqapjprZ/nORN3X1vks9V1bum9h9M8tgk1077OjLJHQv2deUyvCYAWEkvrqpnT9Obkzw/yXu7+4tJUlV/kuQHpuU/nuSkBf3rd1fVg7v771eyYJgXIRBWryOSfKm7T97H8nsWTNc+1lm4/ObuftI+lv/DwRYHAKtFVZ2eWbB7Und/rarek+Tj2ffRvSOSnDp9+QrDcU0grA7/X5J/UlUPqKoHJ/mJJF9L8umq+qnkm9f7Pe4A+3lfkrOr6sjpmr8fm9o/kWRjVT1p2tdRVfWYZXklALDyHprk7ikAPjrJqUkelOR/qqqHTQOf/eSC9f8yyS/snqmqfX3hCuuSEAirQHd/OLPr+m5K8p+TfDTJl5P8dJLzqupvk9yc5KwD7OptST6Z2bWAb0jy19P+v5HkOUl+c9rXjZmdagoA68E7kmyoqluTvDLJB5LsSPJvk3wosy9bb8usb02SFyfZUlU3VdUtSX5+xSuGOarunncNQJLd1yJU1XdldkTv/O7+yLzrAoC1akHfuiGzL0ov6+63zbsumDfXBMLqcUlVnZTZgDCXC4AA/P/t27ENACEMBEG7ty+Mkoip7gNKIADpZipwZq0sc2x091d7t66qmpfngSe4BAIAAATxEwgAABBEBAIAAAQRgQAAAEFEIAAAQBARCAAAEEQEAgAABPkBFM9xzopjy6UAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 1080x720 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aUGFVGiWlLwn"
      },
      "source": [
        "Da questi istogrammi possiamo notare che:\n",
        "\n",
        "\n",
        "* I pazienti sono più donne che uomini\n",
        "* Il numero di pazienti ha due picchi intorno ai 40 anni e agli 80 anni\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 607
        },
        "id": "6ND2Sr53jWAH",
        "outputId": "c5e73c5b-ac86-4fbb-cd05-bdfc6e1f972e"
      },
      "source": [
        "plt.figure(figsize=(20,10))\n",
        "\n",
        "plt.subplot(3,2,1)\n",
        "sb.countplot(x=\"hypertension\", data=stroke_dataset);\n",
        "\n",
        "plt.subplot(3,2,2)\n",
        "sb.countplot(x=\"ever_married\", data=stroke_dataset);\n",
        "\n",
        "plt.subplot(3,2,3)\n",
        "sb.countplot(x=\"smoking_status\", data=stroke_dataset);\n",
        "\n",
        "plt.subplot(3,2,4)\n",
        "sb.countplot(x=\"heart_disease\", data=stroke_dataset);\n",
        "\n",
        "plt.subplot(3,2,5)\n",
        "sb.countplot(x=\"Residence_type\", data=stroke_dataset);\n",
        "\n",
        "plt.subplot(3,2,6)\n",
        "sb.countplot(x=\"work_type\", data=stroke_dataset);\n"
      ],
      "execution_count": 107,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x720 with 6 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "B7XmvMmnmjcy"
      },
      "source": [
        "Da questi altri grafici invece notiamo che:\n",
        "\n",
        "\n",
        "* Sono molti di più i pazienti senza ipertensione che quelli che ne soffrono\n",
        "* Abbiamo meno pazienti che non sono mai stati sposati\n",
        "* Molti dei pazienti non hanno mai fumato, però ce ne sono anche molti in cui abbiamo stato sconosciuto, quindi questo potrebbe andare a pareggiare i conti nella realtà\n",
        "* Pochi pazienti sono cardiopatici\n",
        "* La differenza fra chi abita in zone urbane e chi in zone rurali è pressochè nulla \n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 606
        },
        "id": "ArZJDQdAsYat",
        "outputId": "ec781ae9-f721-4052-e54e-f6aa79dd1c1d"
      },
      "source": [
        "plt.figure(figsize=(20,10))\n",
        "\n",
        "plt.subplot(2,2,1)\n",
        "sb.boxplot(x=\"age\", data=stroke_dataset);\n",
        "\n",
        "plt.subplot(2,2,2)\n",
        "sb.boxplot(x=\"bmi\", data=stroke_dataset);\n",
        "\n",
        "plt.subplot(2,2,3)\n",
        "sb.boxplot(x=\"avg_glucose_level\", data=stroke_dataset);"
      ],
      "execution_count": 108,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAABG0AAAJNCAYAAACLEojLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdf7jed13n+dc7OUkTWoE0KW1JW04zIUM7iyLEWbx0Z5EfY1tdCtcwjq5Oo+MOZZdNM/y4/EEjJRh3XWXLQly9RHRoR0ZlVBQ1qQPojI67IumIgrbi2ZKGBlraFFpK0+ZHP/vHuc/hJDknv3rOuT938nhcV6/c9/e+7+/3k/tzvjnf8+z3e59qrQUAAACAviwZ9gAAAAAAOJ5oAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOjZ3Ok9esWdPGx8cXaCgAwLDdeeedD7XWLhr2OPg6x18AcPab6xjstKLN+Ph4du/ePX+jAgC6UlX3DnsMHM3xFwCc/eY6BnN5FAAAAECHRBsAAACADok2AAAAAB0SbQAAAAA6JNoAAAAAdEi0AQAAAOiQaAMAAADQIdEGAAAAoEOiDQAAAECHRBsAAACADok2AAAAAB0SbQAAAAA6JNoAAAAAdEi0AQAAAOiQaAMAAADQIdEGAAAAoEOiDQAAAECHRBsAAACADok2AAAAAB0SbQAAAAA6JNoAAAAAdEi0AQAAAOjQ2LAHAKdrx44dmZiYGPYwOAX79u1Lkqxdu3bIIzn7rV+/Pps3bx72MABG3mIcZyzm90ffHwBGm2jDyJmYmMinPnNXjjzjwmEPhZNY+vgjSZL7n/RPzUJa+vjDwx4CwFljMY4zFuv7o+8PAKPPT1KMpCPPuDAHXnDdsIfBSay8e2eSmKsFNvU+AzA/Fvo4Y7G+P/r+ADD6fKYNAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6FAX0WbHjh3ZsWPHsIcBAN3zPZP54msJjmafAHo0NuwBJMnExMSwhwAAI8H3TOaLryU4mn0C6FEXZ9oAAAAAcDTRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdGhv2AAAAAHrw6KOP5ju+4zvSWsuyZcty6NChJMnq1auzf//+3Hjjjdm1a1f27t2bsbGxHD58OKtWrcqXv/zl6XVcdtll2b9/f3bs2JFVq1Zl69at+drXvpbPf/7zefvb354PfehDefLJJ3Pffffl4MGDWbduXd74xjdm69atueiii/KlL30prbUkycUXX5z7778/hw4dysUXX5wHHngg4+Pjee1rX5tbb701y5Yty9jYWHbs2JEk2bJlS97znvckSW666aZccsklWbJkSaoqVZWlS5dm+/btWb169XF/9/3792fbtm255ZZbkmT69urVqzMxMZHNmzfn8ssvz4/8yI/kve997/Rjs5m5rpnPmVp+0003nXQdJ1vXQljMbfW4/VExrPdpWNt1pg0AAECSe++9dzqYTAWbZPKHtST5xV/8xezduzdJcvjw4SQ5KtgkyX333ZcDBw5k+/btue2223LXXXdl7969aa3lp37qp3LXXXflnnvuycGDB5Mk99xzT2655ZYcOHAge/fuzRNPPJEnn3wyTz75ZPbu3ZuDBw+mtZb7778/rbV87nOfy7vf/e7pMU5ta/v27fna1742ffvxxx/PPffck4mJifz93/99PvvZz+auu+7K7bffPuvf/bbbbsunP/3p3H777UfdTpLt27fnwIED+exnP5vt27cf9djJ1jXb8lNZx8nWtRAWc1s9bn9UDOt9GtZ2RRsAAOCc9+ijj+app56at/Xt2bMnO3fuPGrZVOg51mOPPXZa654KSzO3tWfPnuNuz2bXrl3TEWrK/v37c8cdd6S1ll27dmXXrl1preWOO+7I7t27j1rfnj17ph87dj3Hrmvmc2YuP9k6TrauhbCY2+px+6NiWO/TMOeni8uj9u3blwMHDmTLli3DHgojYGJiIksOtpM/Ec4RS554NBMTX/Vv6DliYmIiK1euHPYwOAs4/prd2XSc4fvD6fnc5z437+ucK9IM06FDh3L77bfnTW960/Sy2267bTpYzTzD6MiRI3nHO94x63qOHDly3HqOXdfM58xcfrJ1nGxdC2Ext9Xj9kfFsN6nYc7PSc+0qarXV9Xuqtr94IMPLsaYAADOaY6/gIXSWstHP/rRo5Z97GMfmw5MrbXpM3kOHz4851lAhw8fPm49x65r5nNmLj/ZOk62roWwmNvqcfujYljv0zDn56Rn2rTW3pfkfUmycePGBfnfDmvXrk2S6Q/NghPZsmVL7rzngWEPA7rx1IpnZv26i/0beo7wf8zPDY6/hudsOs7w/eH0vPzlL5/Xy6N6VVV51ateddSyV77yldm5c2cOHz6cqkoyGW/GxsayYsWKWcPN2NjYces5dl0znzNz+cnWcbJ1LYTF3FaP2x8Vw3qfhjk/PtMGAAA45z3vec+b93WOjXXxaRRHWbZsWW644Yajlm3atClLliyZfnxq3EuXLp3z8qilS5cet55j1zXzOTOXn2wdJ1vXQljMbfW4/VExrPdpmPMj2gAAAOe8Zz7zmcdFhadjfHw811133VHL5oo4F1xwwWmte+psmJnbGh8fP+72bK699trjfl3x6tWrc80116Sqcu211+baa69NVeWaa67Jxo0bj1rf+Pj49GOz/drjmeua+ZyZy0+2jpOtayEs5rZ63P6oGNb7NMz5EW0AAAAyebbNVBBZtmzZ9PKpH9BuvPHGXHHFFUm+HmBWrVp11Douu+yyrFy5Mlu3bs2mTZty1VVX5YorrkhV5eabb85VV12VdevWZfny5UmSdevWZdu2bVm5cmWuuOKKrFixIuedd17OO++8XHHFFVm+fHmqKpdcckmqKldeeeX0B6AuW7Zseltbt27N+eefP337Gc94RtatW5f169fn+c9/fjZs2JCrrrpqzjMENm3alBe+8IW54YYbjrqdJFu3bs3KlSuzYcOGbN269ajHTrau2ZafyjpOtq6FsJjb6nH7o2JY79OwtlvH/rq4E9m4cWPbvXv3vA9i6vp819tyKqauNT/wgutO/mSGauXdk7/m0lwtrJV378xLfGbBOWOhv2dW1Z2ttY0LsnLOiOOvxbUYxxmL9f3R94fTY58AhmmuYzBn2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANChsWEPIEnWr18/7CEAwEjwPZP54msJjmafAHrURbTZvHnzsIcAACPB90zmi68lOJp9AuiRy6MAAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6NDXsAcCaWPv5wVt69c9jD4CSWPr4/SczVAlv6+MNJLh72MADOGgt9nLFY3x99fwAYfaINI2f9+vXDHgKnaN++w0mStWsdMC6si+0XAPNkMf49Xbzvj74/AIw60YaRs3nz5mEPAQA4SznOAKAnPtMGAAAAoEOiDQAAAECHRBsAAACADok2AAAAAB0SbQAAAAA6JNoAAAAAdEi0AQAAAOiQaAMAAADQIdEGAAAAoEOiDQAAAECHRBsAAACADok2AAAAAB0SbQAAAAA6JNoAAAAAdEi0AQAAAOiQaAMAAADQIdEGAAAAoEOiDQAAAECHRBsAAACADok2AAAAAB0SbQAAAAA6JNoAAAAAdKhaa6f+5KoHk9y7AONYk+ShBVgvi8ccjjbzN/rM4ejrZQ6f11q7aNiD4OsW8PjrbNTLfsTczNFoME+jwTz173TmaNZjsNOKNgulqna31jYOexycOXM42szf6DOHo88cwtNnP+qfORoN5mk0mKf+zcccuTwKAAAAoEOiDQAAAECHeok27xv2AHjazOFoM3+jzxyOPnMIT5/9qH/maDSYp9Fgnvr3tOeoi8+0AQAAAOBovZxpAwAAAMAMog0AAABAh4Yebarqmqr6u6qaqKofG/Z4OLGquryq/riq/raq/qaqtgyWX1hVH62qvx/8uWrYY+XEqmppVf1lVf3+4P6VVfWJwb74G1W1fNhjZG5V9eyq+s2quruq7qqqb7Ufjo6qetPg39DPVNWvVdUK+yCcOscjo8UxR98cU4wGxw59qqpfqaovVdVnZiybdf+pSe8dzNdfV9WLT2UbQ402VbU0yf+d5NokVyf5vqq6ephj4qQOJ3lLa+3qJC9N8sbBnP1Yko+31p6f5OOD+/RtS5K7Ztz/P5K8u7W2PsmXk/zwUEbFqXpPkjtaay9I8k2ZnEv74QioqrVJbkqysbX23yRZmuR7Yx+E0+F4ZLQ45uibY4rOOXbo2geSXHPMsrn2n2uTPH/w3+uT/MKpbGDYZ9r84yQTrbV7WmsHk/x6kuuHPCZOoLX2xdbafx3c/mom/1Ffm8l5u23wtNuSvGY4I+RUVNVlSb4ryfsH9yvJy5P85uAp5rBjVfWsJP8kyS8nSWvtYGvtK7EfjpKxJCuraizJM5J8MfZBOGWOR0aHY46+OaYYKY4dOtRa+5MkDx+zeK795/okt7dJf57k2VV16cm2MexoszbJ52fcv2+wjBFQVeNJvjnJJ5Jc3Fr74uCh+5NcPKRhcWr+ryQ/kuSpwf3VSb7SWjs8uG9f7NuVSR5M8m8Hp5u/v6rOj/1wJLTW9iV5V5K9mTzgeiTJnbEPwhlxPNI9xxx9c0wxAhw7jJy59p8z6h/DjjaMqKq6IMlvJfk3rbVHZz7WJn+PvN8l36mq+u4kX2qt3TnssXDGxpK8OMkvtNa+OcnXcsxpy/bDfg2ua74+kwfKz01yfo4/rRY4BY5H+uaYYyQ4phgBjh1G13zsP8OONvuSXD7j/mWDZXSsqpZl8gDpg6213x4sfmDq1K7Bn18a1vg4qW9L8uqq2pPJSxJfnslrmZ89ON0ysS/27r4k97XWPjG4/5uZPOCyH46GVyb5XGvtwdbaoSS/ncn90j4Ip8HxyEhwzNE/xxSjwbHDaJlr/zmj/jHsaPPJJM8ffOr18kx+mNJHhjwmTmBwHfIvJ7mrtXbrjIc+kmTT4PamJL+72GPj1LTWfry1dllrbTyT+9wftda+P8kfJ3nd4GnmsGOttfuTfL6q/uFg0SuS/G3sh6Nib5KXVtUzBv+mTs2ffRBOkeOR0eCYo3+OKUaGY4fRMtf+85EkNwx+i9RLkzwy4zKqOdXk2TrDU1XXZfJa16VJfqW19lNDHRAnVFXfnuRPk3w6X782+W2ZvI78Q0muSHJvku9prR37gUx0pqpeluStrbXvrqp1mfy/YBcm+cskP9Bae3KY42NuVfWiTH6o4/Ik9yT5oUyGePvhCKiqbUn+RSZ/A85fJvmfMnlNs30QToHjkdHjmKNfjilGg2OHPlXVryV5WZI1SR5IckuS38ks+88guP1cJi9tezzJD7XWdp90G8OONgAAAAAcb9iXRwEAAAAwC9EGAAAAoEOiDQAAAECHRBsAAACADok2AAAAAB0SbQAAABZRVY1X1WfO8LXPrarfnO8xAX0aG/YAAAAAODWttS8ked2wxwEsDmfawDmuqn6nqu6sqr+pqtcPlv1wVX22qv6iqn6pqn5usPyiqvqtqvrk4L9vG+7oAQBG1lhVfbCq7qqq36yqZ1TVnqr636vqU1W1u6peXFV/WFX/X1W9IXl6Z+kAo0e0Af5Va+0lSTYmuamq1ib5iSQvTfJtSV4w47nvSfLu1tq3JPlnSd6/2IMFADhL/MMkP99auyrJo0n+l8Hyva21FyX50yQfyORZNS9Nsm0YgwSGy+VRwE1V9drB7cuT/Msk/7m19nCSVNV/SLJh8Pgrk1xdVVOvfWZVXdBae2wxBwwAcBb4fGvtzwa3fzXJTYPbHxn8+ekkF7TWvprkq1X1ZFU9e7EHCQyXaAPnsKp6WSZDzLe21h6vqv+U5O4kV83xkiVJXtpae2JxRggAcNZqc9x/cvDnUzNuT9338xucY1weBee2ZyX58iDYvCCTp96en+S/r6pVVTWWycugpvzHJJun7lTVixZ1tAAAZ48rqupbB7f/xyT/ZZiDAfok2sC57Y5MfgjeXUl+OsmfJ9mX5H9L8hdJ/izJniSPDJ5/U5KNVfXXVfW3Sd6w6CMGADg7/F2SNw6Ow1Yl+YUhjwfoULV27Fl5wLlu6nNqBmfafDjJr7TWPjzscQEAAJxLnGkDzOYdVfWpJJ9J8rkkvzPk8QAAAJxznGkDAAAA0CFn2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0aOx0nrxmzZo2Pj6+QEMBAIbtzjvvfKi1dtGwxwEAwGlGm/Hx8ezevXuhxgIADFlV3TvsMQAAMMnlUQAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDY8MewCjbsWNHJiYmhrb9ffv2JUnWrl07tDGcqfXr12fz5s3DHgYAAAB0S7R5GiYmJvKpz9yVI8+4cCjbX/r4I0mS+58crWlc+vjDwx4CAAAAdG+0ftrv0JFnXJgDL7huKNteeffOJBna9s/U1LgBAACAuflMGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANChLqLNjh07smPHjmEPAzgD9l8AAICFMTbsASTJxMTEsIcAnCH7LwAAwMLo4kwbAAAAAI4m2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANChsWEPADg77N+/Pz/+4z+ee++9NwcPHsyaNWvy4IMP5vLLL8+WLVvyEz/xE7n00kszNjaWxx57LF/4whdSVXn729+eD37wg9m7d28OHjyY5cuXZ3x8PDfeeGNuvvnmPPXUUzl06FCuvPLKvOY1r8mtt96aJLnxxhtz++2351nPelbuv//+VFXWrl2bhx56KN/wDd+QBx98MK94xSvy8Y9/PMuWLcuaNWvyxS9+MVWVd73rXXnkkUfyzne+c3r8Y2NjOXz48PT9t7zlLbn99tvz4IMPpqrSWjul9+F0njub5cuX5+DBg8ctv+iii/LQQw+lqvLUU09NL1+1alW+/OUvH/XcZcuW5dChQ2c8hiRZsmTJ9HZe97rX5Q/+4A/y1FNPZc2aNfnCF74w599xatuXXHJJ7r///unlS5cuzZEjR6b/nPKc5zwnX/nKV3Lo0KF8//d/f371V381b3nLW7Jr16488cQT+fznPz/9d1mzZk0eeuih3HLLLfnGb/zGvPWtb82ePXuyatWqPPzww1m+fHl+/ud/PuvXr0+STExMZMuWLXnrW9+an/7pn87Bgwdz3nnn5Ud/9Efzrne9K+985zvz/ve/P0eOHMnBgwfzwAMP5NJLL82SJUuyfPnyvPnNb87P/MzPZO/evXnuc5+b8847L9u3b0+SbNu2LbfccktWr149/XfZv3//Uctnu79169ZUVX7yJ39y1te+5jWvyfbt2/OzP/uzeclLXjLre3zsegEAOHvV6fxwsXHjxrZ79+55H8SWLVuSJO95z3vmfd0LacuWLbnzngdy4AXXDWX7K+/emSRD2/6ZWnn3zrxk3cUjN9/Mbmr/fd7znpePfOQjsz7nggsuyGOPPTbrY8fGkhO95ukGkZnrfuKJJ2bd7nxvi9N3svd+bGws11133axfb+Pj4/nABz6QJPnBH/zB7Nmz57ivsan7J/q6nFrXnj17jlp2/fXXp7WW3/u938urX/3qvOlNb5p+7NZbbz1q+Wz3p8Z8/fXXz/rapUuXTo/t93//92cd17HrnW9VdWdrbeO8rxgAgNPm8ijgaTt06FB27do15+Mn+sF4rnAy22vmK6I89thjJww287ktTt/J3vvDhw/PGQj37NmTiYmJTExMTAeXY+d66v6Jvi6n1nWsnTt35o477khrLXfccUf279+fZPLsl5nLJyYmjrs/cx/ZtWvXrK+dObY777zzuO0fu52pdQAAcHbq4vKoffv25cCBA9P/x35UTExMZMlBP9idriVPPJqJia+O3Hwzu4mJiRw+fPhpX44D82XqEqaFcOjQoVRVkuTIkSO5/fbb86Y3vSm33Xbb9OVkR44cyfbt24+7PzMeHTp0aNbXznTLLbccd7bNsduZWgcAAGenk55pU1Wvr6rdVbX7wQcfXIwxASNmts9ggWHZs2fPrGfJzJepM4EOHz6cj370o0mSj33sY9NR5vDhw9mzZ89x92eeQdRam/W1M812JtCx25laBwAAZ6eTnmnTWntfkvclk59psxCDWLt2bZLR/UwbTs9TK56Z9T7T5qyxZcuW3HfffS7ToBvj4+NJZr+8aT5MfebO2NhYXvWqVyVJXvnKV2bnzp05fPhwxsbGctlll+W+++476v699947HW6qatbXznTBBRcct+1jtzO1DgAAzk4+0wZ42i6++OIsW7Zs2MOAJMnWrVuzdevWBVn3smXLpr/Wly5dmhtuuCFJsmnTpixZsmR6+datW4+7PzY2dtR6ZnvtTNu2bTtu2bHbmVoHAABnJ9EGeNqWLVuWa6+9ds7HZztjYMrMH2RP9pqpzxJ5ui644II5tzvf2+L0nVrgNZIAAA2ySURBVOy9Hxsby6tf/epZHxsfH8/69euzfv366TNujp3rqfsn+rqcWtexrrvuulxzzTWpqlxzzTXTv3J79erVRy1fv379cfdn7iPXXnvtrK+dObbZfuX3sdvxK78BAM5uog0wLzZt2pQNGzbkvPPOS1XloosuSpJcfvnlecc73pGVK1dm3bp12bBhQ5773Ocmmfzh/Oabb8769euzfPnyJMny5cuzYcOGbNu2LStWrMjy5ctTVVm3bt1RH7h64403ZuXKlbnkkkum13XZZZdlxYoV09t+xStekWQyKl166aXTz9u2bVve9ra3HTX+Y3+wf/Ob3zy9ntMJOE839ky9D8e66KKLUlXHnZGxatWq4547H2c9zdzO6173uqxcuTLnnXde1q5de8K/49S2p+ZlytKlS4/6c8pznvOc6Tn+gR/4gSST7/3VV1+ddevWHfV3WbNmTZLk5ptvzqZNm3LllVemqnLhhRcmmXzvZp5hs3Xr1px//vm5+eabp78uV6xYkZtvvjnnn39+tm3blquuuiobNmzI+Pj49Nfo+vXrc/XVV2fr1q3ZsGFDVqxYkXXr1uWqq67KDTfckE2bNuWFL3zhcWe5HLt8tvtXXXVVrr766jlf+7a3vS1LliyZ9SybubYDAMDZq07n19pu3Lix7d69e94HMfVbhEbtM06mPtPmwAuuG8r2V969M0mGtv0ztfLunXmJz7Q5a4zq/gvMrqrubK1tHPY4AABwpg0AAABAl0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTaAAAAAHRItAEAAADokGgDAAAA0CHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOjQ27AEkyfr164c9BOAM2X8BAAAWRhfRZvPmzcMeAnCG7L8AAAALw+VRAAAAAB0SbQAAAAA6JNoAAAAAdEi0AQAAAOiQaAMAAADQIdEGAAAAoEOiDQAAAECHRBsAAACADok2AAAAAB0SbQAAAAA6JNoAAAAAdEi0AQAAAOiQaAMAAADQIdEGAAAAoEOiDQAAAECHRBsAAACADok2AAAAAB0SbQAAAAA6JNoAAAAAdEi0AQAAAOiQaAMAAADQIdEGAAAAoEOiDQAAAECHRBsAAACADok2AAAAAB0SbQAAAAA6JNoAAAAAdEi0AQAAAOiQaAMAAADQIdEGAAAAoEOiDQAAAECHRBsAAACADok2AAAAAB0SbQAAAAA6JNoAAAAAdEi0AQAAAOiQaAMAAADQIdEGAAAAoEOiDQAAAECHRBsAAACADok2AAAAAB0SbQAAAAA6JNoAAAAAdEi0AQAAAOiQaAMAAADQIdEGAAAAoENjwx7AqFv6+MNZeffOIW17f5IMbftnaunjDye5eNjDAAAAgK6JNk/D+vXrh7r9ffsOJ0nWrh21AHLx0N87AAAA6J1o8zRs3rx52EMAAAAAzlI+0wYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6JBoAwAAANAh0QYAAACgQ6INAAAAQIdEGwAAAIAOiTYAAAAAHRJtAAAAADok2gAAAAB0SLQBAAAA6FC11k79yVUPJrn3NNa/JslDpzsoFp15Gh3majSYp9Fhro73vNbaRcMeBAAApxltTnvlVbtbaxsXbAPMC/M0OszVaDBPo8NcAQDQM5dHAQAAAHRItAEAAADo0EJHm/ct8PqZH+ZpdJir0WCeRoe5AgCgWwv6mTYAAAAAnBmXRwEAAAB0SLQBAAAA6NC8RZuq2lNVn66qT1XV7sGyC6vqo1X194M/V83X9jh1VfUrVfWlqvrMjGWzzk1Nem9VTVTVX1fVi4c38nPLHPP0jqraN9ivPlVV18147McH8/R3VfWdwxn1uamqLq+qP66qv62qv6mqLYPl9quOnGCe7FcAAIyE+T7T5jtaay9qrW0c3P+xJB9vrT0/yccH91l8H0hyzTHL5pqba5M8f/Df65P8wiKNkdnnKUnePdivXtRa25kkVXV1ku9N8o8Gr/n5qlq6aCPlcJK3tNauTvLSJG8czIn9qi9zzVNivwIAYAQs9OVR1ye5bXD7tiSvWeDtMYvW2p8kefiYxXPNzfVJbm+T/jzJs6vq0sUZ6bltjnmay/VJfr219mRr7XNJJpL84wUbHEdprX2xtfZfB7e/muSuJGtjv+rKCeZpLvYrAAC6Mp/RpiX5j1V1Z1W9frDs4tbaFwe3709y8Txuj6dnrrlZm+TzM553X078Qw4L738dXFLzKzMuMTRPnaiq8STfnOQTsV9165h5SuxXAACMgPmMNt/eWntxJi8DeGNV/ZOZD7bJ3y3u94t3yNx07ReS/IMkL0ryxST/53CHw0xVdUGS30ryb1prj858zH7Vj1nmyX4FAMBImLdo01rbN/jzS0k+nMlTyh+YugRg8OeX5mt7PG1zzc2+JJfPeN5lg2UMQWvtgdbakdbaU0l+KV+/VMM8DVlVLctkCPhga+23B4vtV52ZbZ7sVwAAjIp5iTZVdX5VfcPU7ST/NMlnknwkyabB0zYl+d352B7zYq65+UiSGwa/7ealSR6ZcbkHi+yYzz15bSb3q2Rynr63qs6rqisz+QG3f7HY4ztXVVUl+eUkd7XWbp3xkP2qI3PNk/0KAIBRMTZP67k4yYcnj48zluTft9buqKpPJvlQVf1wknuTfM88bY/TUFW/luRlSdZU1X1Jbkny05l9bnYmuS6TH8D5eJIfWvQBn6PmmKeXVdWLMnmZzZ4kNyZJa+1vqupDSf42k78h542ttSPDGPc56tuS/Mskn66qTw2WvS32q97MNU/fZ78CAGAU1OTHLgAAAADQk4X+ld8AAAAAnAHRBgAAAKBDog0AAABAh0QbAAAAgA6JNgAAAAAdEm0AAAAAOiTawFmuql5WVb8/7HHMVFXjVfWZ3tcJAAAwTKINAAAAQIdEG1hkVfU7VXVnVf1NVb2+qt5QVT874/EfrKqfG9z+iar6u6r6L1X1a1X11hOs91uq6q+r6lNV9bOznXVSVe+YuY6q+kxVjQ9u3zB4/V9V1b8bLBuvqj8aLP94VV0xWP7PB6/9q6r6k8GypYPtfnLw/BtP8f2Y9XVV9etV9V0znveBqnrdmW4HAABg1Ig2sPj+VWvtJUk2JrkpyYeTvHbG4/8iya9X1bck+WdJvinJtYPnn8i/TXJja+1FSY6czoCq6h8l2Zrk5a21b0qyZfDQjiS3tda+MckHk7x3sPztSb5z8NxXD5b9cJJHWmvfkuRbkvzrqrryFDY/1+t+I8n3DMa3PMkrkvzB09gOAADASBFtYPHdVFV/leTPk1ye5Mok91TVS6tqdZIXJPmzJN+W5Hdba0+01r6a5PfmWmFVPTvJN7TW/t/Bon9/mmN6eZL/0Fp7KElaaw8Pln/rjHX9uyTfPrj9Z0k+UFX/OsnSwbJ/muSGqvpUkk8kWZ3k+aew7bletyvJd1TVeZmMVn/SWjvwNLYDAAAwUsaGPQA4l1TVy5K8Msm3ttYer6r/lGRFkl/P5Fkldyf5cGutVdVCDOFwjo61K85kJa21N1TVf5vku5LcWVUvSVJJNrfW/vA0Vzfn6wbvz3dmcPbRiZ4/dZkXAADA2cKZNrC4npXky4Ng84IkLx0s/3CS65N8X74eJ/4syf9QVSuq6oIk3z3XSltrX0ny1UFISZLvneOpe5K8OEmq6sWZPMsnSf4oyT8fnOmTqrpwsPz/mbGu70/yp4PH/0Fr7ROttbcneTCTZwz9YZL/uaqWDZ6zoarOP8n7kZO87jeS/FCS/y7JHafwfAAAgLOGM21gcd2R5A1VdVeSv8vkJVJprX15sOzq1tpfDJZ9sqo+kuSvkzyQ5NNJHjnBun84yS9V1VNJ/vMcz/3/27tDnDqiKAzA/7FNN4DD1JBUswgMS6CmrqlqAgtAoRAYQkJYQEVtE0S7AEgFBofGYiAhB3En4YnHC46h+T414szMzU3G/Dln7s+M0aKrjNGi6+ldV1W1n+RPVT0muUyyk+RbktOq+pERznyZnnNQVZ8yul7Ok/yb1rme5KJGm9Btku1X7MnJivt+Z4xl/eruh1fUAwAA/Dequ996DcALqupjd99V1Yckf5N87e6LVbXT9W6Ste7+vqwWAACA+dNpA/N2XFUbGf+eOXspsJlsVdVexnd9k9EpAwAAwDul0wbemao6yjhZatFhd5++xXpWqarPGeNNi+67e3NZPQAAAM+ENgAAAAAz5PQoAAAAgBkS2gAAAADMkNAGAAAAYIaENgAAAAAz9ARDCu0sxWBnzgAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 1440x720 with 3 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2CxoUvt0huNr"
      },
      "source": [
        "Da questi ultimi grafici possiamo notare che la feature `avg_glucose_level` sembra avere molti valori outlier, a primo impatto potremmo pensare di rimuoverne almeno una parte per non avere valori anomali quindi facciamo un'analisi più approfondita per decidere se è il caso di farlo o no"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 369
        },
        "id": "T1S_VF9Ngxno",
        "outputId": "29ba9d7e-ce39-4514-f938-9459f101574c"
      },
      "source": [
        "sb.displot(x='avg_glucose_level', hue='stroke', data = stroke_dataset, palette=\"pastel\", multiple=\"stack\");"
      ],
      "execution_count": 109,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 402.375x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jD8tze2jifJL",
        "outputId": "72e05dab-1c10-49fa-dd68-701ada33de20"
      },
      "source": [
        "stroke_avg165 = stroke_dataset[stroke_dataset.avg_glucose_level >= 160].loc[:,\"stroke\"].sum()\n",
        "stroke_tot = stroke_dataset.loc[:,\"stroke\"].sum()\n",
        "stroke_perc = stroke_avg165 / stroke_tot\n",
        "print(f\"Abbiamo {stroke_tot} ictus totali, di cui {stroke_avg165} sono avvenuti in pazienti con un valore di glucosio medio nel sangue non nella norma, in percentuale sono quindi il {stroke_perc:.1%}\")"
      ],
      "execution_count": 110,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Abbiamo 209 ictus totali, di cui 76 sono avvenuti in pazienti con un valore di glucosio medio nel sangue non nella norma, in percentuale sono quindi il 36.4%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "38fg_FgwkzBv"
      },
      "source": [
        "Da questa ulteriore analisi possiamo quindi osservare due cose:\n",
        "\n",
        "\n",
        "1.   Una buona fetta degli ictus totali è presente in persone con il livello medio di glucosio nel sangue non nella norma\n",
        "2.   Se consideriamo due frazioni distinte del dataset, nella prima sezione contenente i pazienti con un livello di glucosio medio inferiore a 160 abbiamo che i casi di ictus sono una piccola frazione dei casi totali mentre nella seconda sezione i casi con ictus sono una frazione molto più grande dei casi totali \n",
        "\n",
        "In virtù di questi risultati decidiamo di tenere anche i dati outlier per la feature `avg_glucose_level`\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LotYfgXbuAVW"
      },
      "source": [
        "###Esplorazione dei legami fra le feature"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 369
        },
        "id": "viUx5rFyuK0y",
        "outputId": "69002e33-0fd4-4ee5-cc9e-22714fce7f1b"
      },
      "source": [
        "sb.displot(x='age', hue='stroke', data = stroke_dataset, palette=\"pastel\", multiple=\"stack\");"
      ],
      "execution_count": 111,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 402.375x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mpOWcvApuUAO"
      },
      "source": [
        "Un primo legame interessante è quello fra età e presenza di ictus, possiamo notare come praticamente la totalità degli ictus si sia verificata in pazienti con più di 35-40 anni.\n",
        "In particolare abbiamo un picco nei pazienti più anziani intorno agli 80 anni"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 866
        },
        "id": "fVInUZS30wKq",
        "outputId": "ae8d43f6-7958-4496-b458-2804024ce38e"
      },
      "source": [
        "plt.figure(figsize=(25,20))\n",
        "\n",
        "plt.subplot(4,2,1)\n",
        "sb.countplot(x='hypertension', hue='stroke', data = stroke_dataset, palette=\"pastel\");\n",
        "\n",
        "plt.subplot(4,2,2)\n",
        "sb.countplot(x='heart_disease', hue='stroke', data = stroke_dataset, palette=\"pastel\");\n",
        "\n",
        "plt.subplot(4,2,3)\n",
        "sb.countplot(x='ever_married', hue='stroke', data = stroke_dataset, palette=\"pastel\");\n",
        "\n",
        "plt.subplot(4,2,4)\n",
        "sb.countplot(x='work_type', hue='stroke', data = stroke_dataset, palette=\"pastel\");\n",
        "\n",
        "plt.subplot(4,2,5)\n",
        "sb.countplot(x='Residence_type', hue='stroke', data = stroke_dataset, palette=\"pastel\");\n",
        "\n",
        "plt.subplot(4,2,6)\n",
        "sb.countplot(x='smoking_status', hue='stroke', data = stroke_dataset, palette=\"pastel\");"
      ],
      "execution_count": 112,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1800x1440 with 6 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cOV0GB6-1xjs"
      },
      "source": [
        "Notiamo quindi che:\n",
        "\n",
        "\n",
        "*   Ipertensione e cardiopatia sembrano influire abbastanza sugli ictus\n",
        "*   Molti dei pazienti che hanno presentato ictus erano sposati\n",
        "*   La distribuzione rispetto alla zona di residenza è uniforme, circa metà e metà\n",
        "*   Abitudini scorrette con il fumo possono incidere sulla presenza di ictus\n",
        "*   La maggior parte dei pazienti con ictus lavorava nel settore privato o era un lavoratore autonomo\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 538
        },
        "id": "s5HVYYVBSAiQ",
        "outputId": "3ef9309d-8e45-4967-9f85-a70f4497f068"
      },
      "source": [
        "corr_matrix = stroke_dataset.corr()\n",
        "mask = np.triu(np.ones_like(corr_matrix, dtype=bool))\n",
        "f, ax = plt.subplots(figsize=(11,9))\n",
        "cmap = sb.diverging_palette(230,20,as_cmap=True)\n",
        "sb.heatmap(corr_matrix, mask=mask, cmap=cmap, vmax=.3, center=0, square=True, linewidths=.5, cbar_kws={\"shrink\": .5});\n",
        "\n",
        "#credits: https://seaborn.pydata.org/examples/many_pairwise_correlations.html?highlight=correlation%20matrix"
      ],
      "execution_count": 113,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 792x648 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ipIEq4uUTQi6"
      },
      "source": [
        "Da questa matrice di correlazione possiamo osservare che:\n",
        "\n",
        "\n",
        "*   L'età sembra influire in modo abbastanza uguale su tute le altre feature presenti \n",
        "*   Gli ictus sono correlati in modo più forte con l'età, in minor modo con ipertensione, cardiopatia e glucosio mentre molto meno con il bmi\n",
        "*   È probabile che chi ha un livello medio di glucosio nel sangue più alto abbia anche problemi di ipertensione e/o cardiopatia\n",
        "\n",
        "*Comunque i valori di correlazione non sono molto alti quindi probabilmente sarà necessario usare modelli con feature non lineari*\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "C6SyvQ29WIed"
      },
      "source": [
        "##Feature Engineering"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jfKArvNOX5Uf"
      },
      "source": [
        "Andiamo ora a trasformare le varie feature categoriche, queste verranno splittate creando tante feature quanti sono i possibili valori che potevano assumere\n",
        "\n",
        "\n",
        "*   Ad esempio la feature `Residence_type` poteva assumere valori *Urban* o *Rural* quindi verranno create due nuove feature `Residence_type_Urban` e `Residence_type_Rural`, se una determinata istanza aveva valore Urban ora avrà valore 1 nella rispettiva feature e valore 0 nell'altra\n",
        "\n",
        "In egual modo vengono trasformate tutte le altre feature categoriche\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SwsRfqLFXPAl"
      },
      "source": [
        "categorical_features = [\"gender\", \"work_type\", \"Residence_type\", \"smoking_status\"]\n",
        "stroke_dataset = pd.get_dummies(stroke_dataset, columns=categorical_features, prefix=categorical_features)"
      ],
      "execution_count": 114,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 427
        },
        "id": "O6QwDFydZPzn",
        "outputId": "1c90e2dc-e19e-4c5f-dc16-9fe859d80f6e"
      },
      "source": [
        "stroke_dataset.head(10)"
      ],
      "execution_count": 115,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>age</th>\n",
              "      <th>hypertension</th>\n",
              "      <th>heart_disease</th>\n",
              "      <th>ever_married</th>\n",
              "      <th>avg_glucose_level</th>\n",
              "      <th>bmi</th>\n",
              "      <th>stroke</th>\n",
              "      <th>gender_Female</th>\n",
              "      <th>gender_Male</th>\n",
              "      <th>gender_Other</th>\n",
              "      <th>work_type_Govt_job</th>\n",
              "      <th>work_type_Never_worked</th>\n",
              "      <th>work_type_Private</th>\n",
              "      <th>work_type_Self-employed</th>\n",
              "      <th>work_type_children</th>\n",
              "      <th>Residence_type_Rural</th>\n",
              "      <th>Residence_type_Urban</th>\n",
              "      <th>smoking_status_Unknown</th>\n",
              "      <th>smoking_status_formerly smoked</th>\n",
              "      <th>smoking_status_never smoked</th>\n",
              "      <th>smoking_status_smokes</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>id</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>9046</th>\n",
              "      <td>67.0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>Yes</td>\n",
              "      <td>228.69</td>\n",
              "      <td>36.6</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>31112</th>\n",
              "      <td>80.0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>Yes</td>\n",
              "      <td>105.92</td>\n",
              "      <td>32.5</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>60182</th>\n",
              "      <td>49.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>171.23</td>\n",
              "      <td>34.4</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1665</th>\n",
              "      <td>79.0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>174.12</td>\n",
              "      <td>24.0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>56669</th>\n",
              "      <td>81.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>186.21</td>\n",
              "      <td>29.0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>53882</th>\n",
              "      <td>74.0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>Yes</td>\n",
              "      <td>70.09</td>\n",
              "      <td>27.4</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10434</th>\n",
              "      <td>69.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>No</td>\n",
              "      <td>94.39</td>\n",
              "      <td>22.8</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>60491</th>\n",
              "      <td>78.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>58.57</td>\n",
              "      <td>24.2</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12109</th>\n",
              "      <td>81.0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>80.43</td>\n",
              "      <td>29.7</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12095</th>\n",
              "      <td>61.0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>Yes</td>\n",
              "      <td>120.46</td>\n",
              "      <td>36.8</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "        age  hypertension  ...  smoking_status_never smoked smoking_status_smokes\n",
              "id                         ...                                                   \n",
              "9046   67.0             0  ...                            0                     0\n",
              "31112  80.0             0  ...                            1                     0\n",
              "60182  49.0             0  ...                            0                     1\n",
              "1665   79.0             1  ...                            1                     0\n",
              "56669  81.0             0  ...                            0                     0\n",
              "53882  74.0             1  ...                            1                     0\n",
              "10434  69.0             0  ...                            1                     0\n",
              "60491  78.0             0  ...                            0                     0\n",
              "12109  81.0             1  ...                            1                     0\n",
              "12095  61.0             0  ...                            0                     1\n",
              "\n",
              "[10 rows x 21 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 115
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_yczoCkUZShQ"
      },
      "source": [
        "Un altro aspetto da considerare è che la feature `ever_married` assume valori *Yes* e *No*, quindi andremo a modificarla applicando la seguente trasformazione:\n",
        "* Yes = 1\n",
        "* No = 0"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hpUh-h50Zj3m"
      },
      "source": [
        "stroke_dataset[\"ever_married\"] = np.where(stroke_dataset[\"ever_married\"] == \"Yes\", 1, 0)"
      ],
      "execution_count": 116,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h04z_w0pauJg",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 427
        },
        "outputId": "0526ce90-abf3-4ad1-d51c-113ab197a871"
      },
      "source": [
        "stroke_dataset.head(10)"
      ],
      "execution_count": 117,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>age</th>\n",
              "      <th>hypertension</th>\n",
              "      <th>heart_disease</th>\n",
              "      <th>ever_married</th>\n",
              "      <th>avg_glucose_level</th>\n",
              "      <th>bmi</th>\n",
              "      <th>stroke</th>\n",
              "      <th>gender_Female</th>\n",
              "      <th>gender_Male</th>\n",
              "      <th>gender_Other</th>\n",
              "      <th>work_type_Govt_job</th>\n",
              "      <th>work_type_Never_worked</th>\n",
              "      <th>work_type_Private</th>\n",
              "      <th>work_type_Self-employed</th>\n",
              "      <th>work_type_children</th>\n",
              "      <th>Residence_type_Rural</th>\n",
              "      <th>Residence_type_Urban</th>\n",
              "      <th>smoking_status_Unknown</th>\n",
              "      <th>smoking_status_formerly smoked</th>\n",
              "      <th>smoking_status_never smoked</th>\n",
              "      <th>smoking_status_smokes</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>id</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>9046</th>\n",
              "      <td>67.0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>228.69</td>\n",
              "      <td>36.6</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>31112</th>\n",
              "      <td>80.0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>105.92</td>\n",
              "      <td>32.5</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>60182</th>\n",
              "      <td>49.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>171.23</td>\n",
              "      <td>34.4</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1665</th>\n",
              "      <td>79.0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>174.12</td>\n",
              "      <td>24.0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>56669</th>\n",
              "      <td>81.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>186.21</td>\n",
              "      <td>29.0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>53882</th>\n",
              "      <td>74.0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>70.09</td>\n",
              "      <td>27.4</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10434</th>\n",
              "      <td>69.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>94.39</td>\n",
              "      <td>22.8</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>60491</th>\n",
              "      <td>78.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>58.57</td>\n",
              "      <td>24.2</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12109</th>\n",
              "      <td>81.0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>80.43</td>\n",
              "      <td>29.7</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12095</th>\n",
              "      <td>61.0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>120.46</td>\n",
              "      <td>36.8</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "        age  hypertension  ...  smoking_status_never smoked  smoking_status_smokes\n",
              "id                         ...                                                    \n",
              "9046   67.0             0  ...                            0                      0\n",
              "31112  80.0             0  ...                            1                      0\n",
              "60182  49.0             0  ...                            0                      1\n",
              "1665   79.0             1  ...                            1                      0\n",
              "56669  81.0             0  ...                            0                      0\n",
              "53882  74.0             1  ...                            1                      0\n",
              "10434  69.0             0  ...                            1                      0\n",
              "60491  78.0             0  ...                            0                      0\n",
              "12109  81.0             1  ...                            1                      0\n",
              "12095  61.0             0  ...                            0                      1\n",
              "\n",
              "[10 rows x 21 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 117
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uuXScT43I4Rb"
      },
      "source": [
        "In questo modo abbiamo ottenuto tutte feature numeriche su cui possiamo lavorare agilmente"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VARNXETCrpIc"
      },
      "source": [
        "Applichiamo una tecnica di oversampling al dataset per ottenere un bilanciamento delle classi"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "t8XRsECEX6PO",
        "outputId": "781e4693-9ce7-420f-dd04-dc42a4941622"
      },
      "source": [
        "sm = SMOTE(random_state=42)\n",
        "y = stroke_dataset[\"stroke\"]\n",
        "X = stroke_dataset.drop(\"stroke\", axis=1)\n",
        "X_res, y_res = sm.fit_resample(X, y)"
      ],
      "execution_count": 118,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/sklearn/utils/deprecation.py:87: FutureWarning: Function safe_indexing is deprecated; safe_indexing is deprecated in version 0.22 and will be removed in version 0.24.\n",
            "  warnings.warn(msg, category=FutureWarning)\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2Z5Ob17QK6EZ"
      },
      "source": [
        "Dal seguente grafico notiamo che ora le classi sono bilanciate"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 264
        },
        "id": "fHdsrY04KIyf",
        "outputId": "0d590639-4c5a-476d-95bf-12a566151368"
      },
      "source": [
        "pd.value_counts(y_res).plot.pie(autopct=\"%.1f%%\", title=\"Stroke\");"
      ],
      "execution_count": 119,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jp78YFGUKbKK"
      },
      "source": [
        "Andiamo ora a suddividere i dati in training set e validations set"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8gg2a3pvKhrm"
      },
      "source": [
        "X_train, X_val, y_train, y_val = train_test_split(\n",
        "    X_res,y_res,\n",
        "    test_size = 1/3,\n",
        "    random_state = 42\n",
        ")"
      ],
      "execution_count": 120,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6mV65CxObBY6"
      },
      "source": [
        "Applichiamo una standardiddazione dei dati"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-soNIeU1bE5n"
      },
      "source": [
        "scaler = StandardScaler()\n",
        "X_train_S = scaler.fit_transform(X_train)\n",
        "X_val_S = scaler.transform(X_val)"
      ],
      "execution_count": 121,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7I3o_5ePHRDZ"
      },
      "source": [
        "In tutti i modelli in cui verrà applicata la `Grid Search` useremo una divisione con la classe `StratifiedKFold` in modo da avere in ogni sub-fold la stessa distribuzione per i dati"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XUne0_O4_0To"
      },
      "source": [
        "La seguente funzione calcola l'intervallo di accuratezza per un modello di classificazione con confidenza al 95% dato l'f-1 score secondo la seguente formula: \n",
        "$$ p = \\frac{f + \\frac{z^2}{2N} \\pm z \\sqrt{\\frac{f}{N} - \\frac{f^2}{N} + \\frac{z^2}{4N^2}  }  }{1+\\frac{z^2}{N}}  $$\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7sbUjYcJ9-Jz"
      },
      "source": [
        "def accuracy_interval(f):\n",
        "  N = len(y_val)\n",
        "  n_min = f + ( 1.96**2/(2*N) - 1.96 * np.sqrt( (f/N) - (f**2/N) + (1.96**2/(4*N**2) ) ) ) \n",
        "  n_max = f + ( 1.96**2/(2*N) + 1.96 * np.sqrt( (f/N) - (f**2/N) + (1.96**2/(4*N**2) ) ) ) \n",
        "  d = 1 + (1.96**2 / N)\n",
        "  e_min = n_min / d\n",
        "  e_max = n_max / d\n",
        "  return np.round(e_min,4), np.round(e_max,4)"
      ],
      "execution_count": 122,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-2Nr-MhZNe07"
      },
      "source": [
        "Creo un dizionario vuoto in cui man mano inserirò i valori f1-score di ogni modello in modo da poter fare un confronto finale"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pVQBkYkxIzb8"
      },
      "source": [
        "accuracy = {}"
      ],
      "execution_count": 123,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zvd-onAwJ7Bt"
      },
      "source": [
        "##Perceptron"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "B-aUZ8s1bvzl",
        "outputId": "baece3c9-bb7f-49f7-c81c-adb235ab4260"
      },
      "source": [
        "model = Perceptron(random_state=42)\n",
        "model.fit(X_train_S, y_train)"
      ],
      "execution_count": 124,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Perceptron(alpha=0.0001, class_weight=None, early_stopping=False, eta0=1.0,\n",
              "           fit_intercept=True, max_iter=1000, n_iter_no_change=5, n_jobs=None,\n",
              "           penalty=None, random_state=42, shuffle=True, tol=0.001,\n",
              "           validation_fraction=0.1, verbose=0, warm_start=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 124
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 111
        },
        "id": "uU5zLXxub9EF",
        "outputId": "3ca17980-0a3a-4ca1-f167-5f101f414e3f"
      },
      "source": [
        "y_pred = model.predict(X_val_S)\n",
        "cm = confusion_matrix(y_val, y_pred)\n",
        "pd.DataFrame(cm, index=model.classes_, columns=model.classes_)"
      ],
      "execution_count": 125,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>0</th>\n",
              "      <th>1</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1170</td>\n",
              "      <td>417</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>389</td>\n",
              "      <td>1158</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      0     1\n",
              "0  1170   417\n",
              "1   389  1158"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 125
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GR4VXd8ndsNc",
        "outputId": "d55f7c30-7163-4cb8-9ccf-fde8f8218dd1"
      },
      "source": [
        "precision = precision_score(y_val, y_pred, pos_label=0)\n",
        "recall = recall_score(y_val, y_pred, pos_label=0)\n",
        "f1 = f1_score(y_val, y_pred, average=\"macro\")\n",
        "\n",
        "print(f\"precision: {precision}, recall: {recall}, f1-score: {f1}\")"
      ],
      "execution_count": 126,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "precision: 0.7504810776138551, recall: 0.7372400756143668, f1-score: 0.7428169058825398\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ykBfnBLN-RuV",
        "outputId": "5376b47e-094d-469a-9a4b-b2c601a59157"
      },
      "source": [
        "i = accuracy_interval(f1)\n",
        "accuracy[\"Perceptron\"] = i\n",
        "print(i)"
      ],
      "execution_count": 127,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(0.7272, 0.7578)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uLrOSKXKV1zG"
      },
      "source": [
        "Aggiungiamo GridSearch e CrossValidation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "RGjBRbBjRDSC",
        "outputId": "cf845d8f-38b2-4c9f-fc66-ba7e8ac36837"
      },
      "source": [
        "model = Perceptron(random_state=42)\n",
        "parameters = {\n",
        "    \"penalty\": [None, \"l1\", \"l2\", \"elasticnet\"],\n",
        "    \"alpha\": np.logspace(-4, 0, 5),\n",
        "    \"tol\": np.logspace(-9, 6, 6)\n",
        "}\n",
        "\n",
        "skf = StratifiedKFold(3, shuffle=True, random_state=42)\n",
        "gs = GridSearchCV(model, parameters, cv=skf)\n",
        "\n",
        "gs.fit(X_train_S, y_train)"
      ],
      "execution_count": 128,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "GridSearchCV(cv=StratifiedKFold(n_splits=3, random_state=42, shuffle=True),\n",
              "             error_score=nan,\n",
              "             estimator=Perceptron(alpha=0.0001, class_weight=None,\n",
              "                                  early_stopping=False, eta0=1.0,\n",
              "                                  fit_intercept=True, max_iter=1000,\n",
              "                                  n_iter_no_change=5, n_jobs=None, penalty=None,\n",
              "                                  random_state=42, shuffle=True, tol=0.001,\n",
              "                                  validation_fraction=0.1, verbose=0,\n",
              "                                  warm_start=False),\n",
              "             iid='deprecated', n_jobs=None,\n",
              "             param_grid={'alpha': array([1.e-04, 1.e-03, 1.e-02, 1.e-01, 1.e+00]),\n",
              "                         'penalty': [None, 'l1', 'l2', 'elasticnet'],\n",
              "                         'tol': array([1.e-09, 1.e-06, 1.e-03, 1.e+00, 1.e+03, 1.e+06])},\n",
              "             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
              "             scoring=None, verbose=0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 128
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 564
        },
        "id": "ECrHP6uTVuGz",
        "outputId": "6d13b9c0-0518-49b0-a491-254d2062027f"
      },
      "source": [
        "pd.DataFrame(gs.cv_results_).sort_values(\"rank_test_score\").head(5)"
      ],
      "execution_count": 129,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>mean_fit_time</th>\n",
              "      <th>std_fit_time</th>\n",
              "      <th>mean_score_time</th>\n",
              "      <th>std_score_time</th>\n",
              "      <th>param_alpha</th>\n",
              "      <th>param_penalty</th>\n",
              "      <th>param_tol</th>\n",
              "      <th>params</th>\n",
              "      <th>split0_test_score</th>\n",
              "      <th>split1_test_score</th>\n",
              "      <th>split2_test_score</th>\n",
              "      <th>mean_test_score</th>\n",
              "      <th>std_test_score</th>\n",
              "      <th>rank_test_score</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>30</th>\n",
              "      <td>0.019622</td>\n",
              "      <td>0.004117</td>\n",
              "      <td>0.001059</td>\n",
              "      <td>0.000030</td>\n",
              "      <td>0.001</td>\n",
              "      <td>l1</td>\n",
              "      <td>1e-09</td>\n",
              "      <td>{'alpha': 0.001, 'penalty': 'l1', 'tol': 1e-09}</td>\n",
              "      <td>0.757300</td>\n",
              "      <td>0.780756</td>\n",
              "      <td>0.742816</td>\n",
              "      <td>0.760291</td>\n",
              "      <td>0.015633</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>31</th>\n",
              "      <td>0.023239</td>\n",
              "      <td>0.004066</td>\n",
              "      <td>0.002132</td>\n",
              "      <td>0.001516</td>\n",
              "      <td>0.001</td>\n",
              "      <td>l1</td>\n",
              "      <td>1e-06</td>\n",
              "      <td>{'alpha': 0.001, 'penalty': 'l1', 'tol': 1e-06}</td>\n",
              "      <td>0.757300</td>\n",
              "      <td>0.780756</td>\n",
              "      <td>0.742816</td>\n",
              "      <td>0.760291</td>\n",
              "      <td>0.015633</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>32</th>\n",
              "      <td>0.021980</td>\n",
              "      <td>0.007194</td>\n",
              "      <td>0.001096</td>\n",
              "      <td>0.000072</td>\n",
              "      <td>0.001</td>\n",
              "      <td>l1</td>\n",
              "      <td>0.001</td>\n",
              "      <td>{'alpha': 0.001, 'penalty': 'l1', 'tol': 0.001}</td>\n",
              "      <td>0.757300</td>\n",
              "      <td>0.780756</td>\n",
              "      <td>0.742816</td>\n",
              "      <td>0.760291</td>\n",
              "      <td>0.015633</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>47</th>\n",
              "      <td>0.012465</td>\n",
              "      <td>0.002032</td>\n",
              "      <td>0.001132</td>\n",
              "      <td>0.000146</td>\n",
              "      <td>0.001</td>\n",
              "      <td>elasticnet</td>\n",
              "      <td>1e+06</td>\n",
              "      <td>{'alpha': 0.001, 'penalty': 'elasticnet', 'tol...</td>\n",
              "      <td>0.745333</td>\n",
              "      <td>0.732408</td>\n",
              "      <td>0.760057</td>\n",
              "      <td>0.745933</td>\n",
              "      <td>0.011296</td>\n",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>46</th>\n",
              "      <td>0.010388</td>\n",
              "      <td>0.000166</td>\n",
              "      <td>0.001123</td>\n",
              "      <td>0.000136</td>\n",
              "      <td>0.001</td>\n",
              "      <td>elasticnet</td>\n",
              "      <td>1000</td>\n",
              "      <td>{'alpha': 0.001, 'penalty': 'elasticnet', 'tol...</td>\n",
              "      <td>0.745333</td>\n",
              "      <td>0.732408</td>\n",
              "      <td>0.760057</td>\n",
              "      <td>0.745933</td>\n",
              "      <td>0.011296</td>\n",
              "      <td>4</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "    mean_fit_time  std_fit_time  ...  std_test_score  rank_test_score\n",
              "30       0.019622      0.004117  ...        0.015633                1\n",
              "31       0.023239      0.004066  ...        0.015633                1\n",
              "32       0.021980      0.007194  ...        0.015633                1\n",
              "47       0.012465      0.002032  ...        0.011296                4\n",
              "46       0.010388      0.000166  ...        0.011296                4\n",
              "\n",
              "[5 rows x 14 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 129
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WtTCXQ2HU0ck",
        "outputId": "43bfcb24-fb53-4212-bd57-6588b681b56d"
      },
      "source": [
        "gs.best_params_"
      ],
      "execution_count": 130,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'alpha': 0.001, 'penalty': 'l1', 'tol': 1e-09}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 130
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "oCDPFWtzU3RQ",
        "outputId": "c609e752-00e5-4c1c-cff1-3b300a6800da"
      },
      "source": [
        "print(f'Best score: {round(gs.best_score_ * 100, 4):.4f}%')"
      ],
      "execution_count": 131,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Best score: 76.0291%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "c5XXuQ1HWEfo",
        "outputId": "02866cf0-cc9c-4f60-f9e8-7fa7eac502ff"
      },
      "source": [
        "pc_imp = pd.Series(gs.best_estimator_.coef_[0], index=X.columns)\n",
        "pc_imp.nlargest(4).plot(kind='barh');"
      ],
      "execution_count": 132,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "at1Wh_lTW5Xh"
      },
      "source": [
        "A seguito della penalizzazione l1, che la GridSearch identifica come parametro migliore, notiamo che le feature più rilevanti sono:\n",
        "\n",
        "\n",
        "*   Età\n",
        "*   Tipo di lavoro: privato\n",
        "*   Livello medio di glucosio nel sangue\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 111
        },
        "id": "Ms8qSIF6XSdt",
        "outputId": "f5370775-a22d-41ef-adc8-9e0607a3050d"
      },
      "source": [
        "y_pred = gs.predict(X_val_S)\n",
        "cm = confusion_matrix(y_val, y_pred)\n",
        "pd.DataFrame(cm, index=gs.best_estimator_.classes_, columns=gs.best_estimator_.classes_)"
      ],
      "execution_count": 133,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>0</th>\n",
              "      <th>1</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1315</td>\n",
              "      <td>272</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>655</td>\n",
              "      <td>892</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      0    1\n",
              "0  1315  272\n",
              "1   655  892"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 133
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "n_lOJ3YrXr4f",
        "outputId": "9ad0a6ec-255b-477e-b3f9-4ef9d47005da"
      },
      "source": [
        "precision = precision_score(y_val, y_pred, pos_label=0)\n",
        "recall = recall_score(y_val, y_pred, pos_label=0)\n",
        "f1 = f1_score(y_val, y_pred, average=\"macro\")\n",
        "\n",
        "print(f\"precision: {precision}, recall: {recall}, f1-score: {f1}\")"
      ],
      "execution_count": 134,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "precision: 0.6675126903553299, recall: 0.8286074354127284, f1-score: 0.6987234402641411\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "9uDLOFq8-qas",
        "outputId": "d1eea8a8-a949-4791-c3ee-e828d38938b1"
      },
      "source": [
        "i = accuracy_interval(f1)\n",
        "accuracy[\"Perceptron with gs\"] = i\n",
        "print(i)"
      ],
      "execution_count": 135,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(0.6824, 0.7145)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rPKAqNvxbFeh",
        "outputId": "2d4bd35b-ea57-4c24-8d25-cf5d87d23476"
      },
      "source": [
        "perceptron_mse = mean_squared_error(y_val, y_pred)\n",
        "print('MSE: {}'.format(perceptron_mse))"
      ],
      "execution_count": 136,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "MSE: 0.295788130185067\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "G73ReVsRKApL"
      },
      "source": [
        "##Logistic regression"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "P13-R9udfbQ1"
      },
      "source": [
        "model = LogisticRegression(random_state=42, solver=\"saga\")"
      ],
      "execution_count": 137,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "E3IVo4wgfgac",
        "outputId": "c3373031-e994-489a-8854-415921fa6b63"
      },
      "source": [
        "model.fit(X_train_S, y_train)"
      ],
      "execution_count": 138,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
              "                   intercept_scaling=1, l1_ratio=None, max_iter=100,\n",
              "                   multi_class='auto', n_jobs=None, penalty='l2',\n",
              "                   random_state=42, solver='saga', tol=0.0001, verbose=0,\n",
              "                   warm_start=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 138
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 111
        },
        "id": "HZpp0gVSfpOl",
        "outputId": "1cc03809-f839-40d9-b4e7-db954fbd6020"
      },
      "source": [
        "y_pred = model.predict(X_val_S)\n",
        "cm = confusion_matrix(y_val, y_pred)\n",
        "pd.DataFrame(cm, index=model.classes_, columns=model.classes_)"
      ],
      "execution_count": 139,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>0</th>\n",
              "      <th>1</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1210</td>\n",
              "      <td>377</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>295</td>\n",
              "      <td>1252</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      0     1\n",
              "0  1210   377\n",
              "1   295  1252"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 139
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pYn44J5jfvw9",
        "outputId": "3ed4384f-016d-45bd-ecb2-35a43052648e"
      },
      "source": [
        "precision = precision_score(y_val, y_pred, pos_label=0)\n",
        "recall = recall_score(y_val, y_pred, pos_label=0)\n",
        "f1 = f1_score(y_val, y_pred, average=\"macro\")\n",
        "\n",
        "print(f\"precision: {precision}, recall: {recall}, f1-score: {f1}\")"
      ],
      "execution_count": 140,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "precision: 0.8039867109634552, recall: 0.7624448645242596, f1-score: 0.785539020011014\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Df6oG0dHJObV",
        "outputId": "158986c9-6029-4bae-bcda-0d502e3d7fca"
      },
      "source": [
        "i = accuracy_interval(f1)\n",
        "accuracy[\"Logistic Regression\"] = i\n",
        "print(i)"
      ],
      "execution_count": 141,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(0.7708, 0.7996)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WVYHV5CRYWlv",
        "outputId": "a399ebe2-a3bd-4f66-d93d-897a48af29c2"
      },
      "source": [
        "model = LogisticRegression(random_state=42, solver=\"saga\")\n",
        "parameters = {\n",
        "    \"penalty\": [\"l1\"],\n",
        "    \"C\": [0.3, 0.8, 1],#np.logspace(-4, 0, 5),\n",
        "    \"tol\": np.logspace(-9, 6, 6)\n",
        "}\n",
        "\n",
        "skf = StratifiedKFold(3, shuffle=True, random_state=42)\n",
        "gs = GridSearchCV(model, parameters, cv=skf)\n",
        "\n",
        "gs.fit(X_train_S, y_train)"
      ],
      "execution_count": 142,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_sag.py:330: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge\n",
            "  \"the coef_ did not converge\", ConvergenceWarning)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "GridSearchCV(cv=StratifiedKFold(n_splits=3, random_state=42, shuffle=True),\n",
              "             error_score=nan,\n",
              "             estimator=LogisticRegression(C=1.0, class_weight=None, dual=False,\n",
              "                                          fit_intercept=True,\n",
              "                                          intercept_scaling=1, l1_ratio=None,\n",
              "                                          max_iter=100, multi_class='auto',\n",
              "                                          n_jobs=None, penalty='l2',\n",
              "                                          random_state=42, solver='saga',\n",
              "                                          tol=0.0001, verbose=0,\n",
              "                                          warm_start=False),\n",
              "             iid='deprecated', n_jobs=None,\n",
              "             param_grid={'C': [0.3, 0.8, 1], 'penalty': ['l1'],\n",
              "                         'tol': array([1.e-09, 1.e-06, 1.e-03, 1.e+00, 1.e+03, 1.e+06])},\n",
              "             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
              "             scoring=None, verbose=0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 142
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 479
        },
        "id": "mf0_XF_3ZQi-",
        "outputId": "20e92278-6103-462d-d44b-4be11785ff50"
      },
      "source": [
        "pd.DataFrame(gs.cv_results_).sort_values(\"rank_test_score\").head(5)"
      ],
      "execution_count": 143,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>mean_fit_time</th>\n",
              "      <th>std_fit_time</th>\n",
              "      <th>mean_score_time</th>\n",
              "      <th>std_score_time</th>\n",
              "      <th>param_C</th>\n",
              "      <th>param_penalty</th>\n",
              "      <th>param_tol</th>\n",
              "      <th>params</th>\n",
              "      <th>split0_test_score</th>\n",
              "      <th>split1_test_score</th>\n",
              "      <th>split2_test_score</th>\n",
              "      <th>mean_test_score</th>\n",
              "      <th>std_test_score</th>\n",
              "      <th>rank_test_score</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>0.041481</td>\n",
              "      <td>0.004075</td>\n",
              "      <td>0.002396</td>\n",
              "      <td>0.001940</td>\n",
              "      <td>0.8</td>\n",
              "      <td>l1</td>\n",
              "      <td>0.001</td>\n",
              "      <td>{'C': 0.8, 'penalty': 'l1', 'tol': 0.001}</td>\n",
              "      <td>0.791766</td>\n",
              "      <td>0.791288</td>\n",
              "      <td>0.765805</td>\n",
              "      <td>0.782953</td>\n",
              "      <td>0.012127</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>0.200596</td>\n",
              "      <td>0.003740</td>\n",
              "      <td>0.002223</td>\n",
              "      <td>0.001305</td>\n",
              "      <td>0.8</td>\n",
              "      <td>l1</td>\n",
              "      <td>1e-06</td>\n",
              "      <td>{'C': 0.8, 'penalty': 'l1', 'tol': 1e-06}</td>\n",
              "      <td>0.791288</td>\n",
              "      <td>0.791288</td>\n",
              "      <td>0.765805</td>\n",
              "      <td>0.782793</td>\n",
              "      <td>0.012013</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>0.202260</td>\n",
              "      <td>0.001842</td>\n",
              "      <td>0.002071</td>\n",
              "      <td>0.001162</td>\n",
              "      <td>0.8</td>\n",
              "      <td>l1</td>\n",
              "      <td>1e-09</td>\n",
              "      <td>{'C': 0.8, 'penalty': 'l1', 'tol': 1e-09}</td>\n",
              "      <td>0.791288</td>\n",
              "      <td>0.791288</td>\n",
              "      <td>0.765805</td>\n",
              "      <td>0.782793</td>\n",
              "      <td>0.012013</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>0.040008</td>\n",
              "      <td>0.004562</td>\n",
              "      <td>0.000992</td>\n",
              "      <td>0.000012</td>\n",
              "      <td>1</td>\n",
              "      <td>l1</td>\n",
              "      <td>0.001</td>\n",
              "      <td>{'C': 1, 'penalty': 'l1', 'tol': 0.001}</td>\n",
              "      <td>0.791766</td>\n",
              "      <td>0.791288</td>\n",
              "      <td>0.765326</td>\n",
              "      <td>0.782793</td>\n",
              "      <td>0.012353</td>\n",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>0.203524</td>\n",
              "      <td>0.004758</td>\n",
              "      <td>0.004655</td>\n",
              "      <td>0.002174</td>\n",
              "      <td>1</td>\n",
              "      <td>l1</td>\n",
              "      <td>1e-06</td>\n",
              "      <td>{'C': 1, 'penalty': 'l1', 'tol': 1e-06}</td>\n",
              "      <td>0.791766</td>\n",
              "      <td>0.791288</td>\n",
              "      <td>0.765326</td>\n",
              "      <td>0.782793</td>\n",
              "      <td>0.012353</td>\n",
              "      <td>4</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "    mean_fit_time  std_fit_time  ...  std_test_score  rank_test_score\n",
              "8        0.041481      0.004075  ...        0.012127                1\n",
              "7        0.200596      0.003740  ...        0.012013                2\n",
              "6        0.202260      0.001842  ...        0.012013                2\n",
              "14       0.040008      0.004562  ...        0.012353                4\n",
              "13       0.203524      0.004758  ...        0.012353                4\n",
              "\n",
              "[5 rows x 14 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 143
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "SqVH-Kl1ZVPf",
        "outputId": "6f9b26eb-8c1f-4148-fe63-8eab96c99303"
      },
      "source": [
        "gs.best_params_"
      ],
      "execution_count": 144,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'C': 0.8, 'penalty': 'l1', 'tol': 0.001}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 144
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gvau57tpZZ-b",
        "outputId": "4c0fee0d-7390-4918-fe56-c399c5184504"
      },
      "source": [
        "print(f'Best score: {round(gs.best_score_ * 100, 4):.4f}%')"
      ],
      "execution_count": 145,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Best score: 78.2953%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "ylFC3gVlZr-Q",
        "outputId": "d9a39b3d-0742-4df3-888e-59982842273e"
      },
      "source": [
        "pc_imp = pd.Series(gs.best_estimator_.coef_[0], index=X.columns)\n",
        "pc_imp.nlargest(4).plot(kind='barh');"
      ],
      "execution_count": 146,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 111
        },
        "id": "zvBU5Y77Zy7H",
        "outputId": "74febd64-a7ce-4558-90d2-ed4ab01dd9f8"
      },
      "source": [
        "y_pred = gs.predict(X_val_S)\n",
        "cm = confusion_matrix(y_val, y_pred)\n",
        "pd.DataFrame(cm, index=gs.best_estimator_.classes_, columns=gs.best_estimator_.classes_)"
      ],
      "execution_count": 147,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>0</th>\n",
              "      <th>1</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1209</td>\n",
              "      <td>378</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>294</td>\n",
              "      <td>1253</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      0     1\n",
              "0  1209   378\n",
              "1   294  1253"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 147
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "PvS0TlieZ00B",
        "outputId": "34543cf4-9e27-4080-95e7-2c10d6c7031f"
      },
      "source": [
        "precision = precision_score(y_val, y_pred, pos_label=0)\n",
        "recall = recall_score(y_val, y_pred, pos_label=0)\n",
        "f1 = f1_score(y_val, y_pred, average=\"macro\")\n",
        "\n",
        "print(f\"precision: {precision}, recall: {recall}, f1-score: {f1}\")"
      ],
      "execution_count": 148,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "precision: 0.8043912175648703, recall: 0.7618147448015122, f1-score: 0.7855352636756341\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5pqGm9_p-zW6",
        "outputId": "8e7ff085-03c3-4242-be70-36705a0d096c"
      },
      "source": [
        "i = accuracy_interval(f1)\n",
        "accuracy[\"Logistic regression with gs\"] = i\n",
        "print(i)"
      ],
      "execution_count": 149,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(0.7708, 0.7996)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XXCdBcoObSBY",
        "outputId": "c3af2b00-1e3c-415f-9727-145ee428ba19"
      },
      "source": [
        "logisticregression_mse = mean_squared_error(y_val, y_pred)\n",
        "print('MSE: {}'.format(logisticregression_mse))"
      ],
      "execution_count": 150,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "MSE: 0.21442246330567966\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ItRC6PZ1ySJk"
      },
      "source": [
        "##SVM"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YZi_QGLuyvPt",
        "outputId": "d9d7121a-c6fe-43fe-8722-2da1d26336eb"
      },
      "source": [
        "model = SVC()\n",
        "parameters = {\n",
        "    \"kernel\": [\"rbf\"],\n",
        "    \"C\": np.logspace(-2, 0, 3)\n",
        "}\n",
        "\n",
        "skf = StratifiedKFold(3, shuffle=True, random_state=42)\n",
        "gs = GridSearchCV(model, parameters, cv=skf)\n",
        "\n",
        "gs.fit(X_train_S, y_train)"
      ],
      "execution_count": 151,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "GridSearchCV(cv=StratifiedKFold(n_splits=3, random_state=42, shuffle=True),\n",
              "             error_score=nan,\n",
              "             estimator=SVC(C=1.0, break_ties=False, cache_size=200,\n",
              "                           class_weight=None, coef0=0.0,\n",
              "                           decision_function_shape='ovr', degree=3,\n",
              "                           gamma='scale', kernel='rbf', max_iter=-1,\n",
              "                           probability=False, random_state=None, shrinking=True,\n",
              "                           tol=0.001, verbose=False),\n",
              "             iid='deprecated', n_jobs=None,\n",
              "             param_grid={'C': array([0.01, 0.1 , 1.  ]), 'kernel': ['rbf']},\n",
              "             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
              "             scoring=None, verbose=0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 151
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 315
        },
        "id": "2hvSycplzWtd",
        "outputId": "dbe5c9f6-da0b-4b6a-eb82-45f4ea72fc99"
      },
      "source": [
        "pd.DataFrame(gs.cv_results_).sort_values(\"rank_test_score\").head(5)"
      ],
      "execution_count": 152,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>mean_fit_time</th>\n",
              "      <th>std_fit_time</th>\n",
              "      <th>mean_score_time</th>\n",
              "      <th>std_score_time</th>\n",
              "      <th>param_C</th>\n",
              "      <th>param_kernel</th>\n",
              "      <th>params</th>\n",
              "      <th>split0_test_score</th>\n",
              "      <th>split1_test_score</th>\n",
              "      <th>split2_test_score</th>\n",
              "      <th>mean_test_score</th>\n",
              "      <th>std_test_score</th>\n",
              "      <th>rank_test_score</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.405385</td>\n",
              "      <td>0.006832</td>\n",
              "      <td>0.141135</td>\n",
              "      <td>0.001930</td>\n",
              "      <td>1</td>\n",
              "      <td>rbf</td>\n",
              "      <td>{'C': 1.0, 'kernel': 'rbf'}</td>\n",
              "      <td>0.933940</td>\n",
              "      <td>0.923887</td>\n",
              "      <td>0.924330</td>\n",
              "      <td>0.927385</td>\n",
              "      <td>0.004638</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.676020</td>\n",
              "      <td>0.007561</td>\n",
              "      <td>0.238861</td>\n",
              "      <td>0.003836</td>\n",
              "      <td>0.1</td>\n",
              "      <td>rbf</td>\n",
              "      <td>{'C': 0.1, 'kernel': 'rbf'}</td>\n",
              "      <td>0.879847</td>\n",
              "      <td>0.885112</td>\n",
              "      <td>0.861590</td>\n",
              "      <td>0.875516</td>\n",
              "      <td>0.010079</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1.026985</td>\n",
              "      <td>0.032415</td>\n",
              "      <td>0.365636</td>\n",
              "      <td>0.003265</td>\n",
              "      <td>0.01</td>\n",
              "      <td>rbf</td>\n",
              "      <td>{'C': 0.01, 'kernel': 'rbf'}</td>\n",
              "      <td>0.785543</td>\n",
              "      <td>0.788894</td>\n",
              "      <td>0.780172</td>\n",
              "      <td>0.784870</td>\n",
              "      <td>0.003592</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   mean_fit_time  std_fit_time  ...  std_test_score  rank_test_score\n",
              "2       0.405385      0.006832  ...        0.004638                1\n",
              "1       0.676020      0.007561  ...        0.010079                2\n",
              "0       1.026985      0.032415  ...        0.003592                3\n",
              "\n",
              "[3 rows x 13 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 152
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "UjAN1TnpzXG_",
        "outputId": "cd1dc283-c467-42e5-ebca-26ec1f78bac4"
      },
      "source": [
        "gs.best_params_"
      ],
      "execution_count": 153,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'C': 1.0, 'kernel': 'rbf'}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 153
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wz_qTNc3zZuy",
        "outputId": "6b188f0a-ca61-484e-b1ef-c548e1e4c7d6"
      },
      "source": [
        "print(f'Best score: {round(gs.best_score_ * 100, 4):.4f}%')"
      ],
      "execution_count": 154,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Best score: 92.7385%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "BOp5uncX2pz2",
        "outputId": "395f75a2-484a-47b4-e8b2-fa345a9fcec2"
      },
      "source": [
        "svm_imp = pd.Series(gs.best_estimator_.support_vectors_[0], index=X.columns)\n",
        "svm_imp.nlargest(4).plot(kind='barh');"
      ],
      "execution_count": 155,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 111
        },
        "id": "tbjeZyU_z0gs",
        "outputId": "4ac391d5-7c03-45d2-8e25-25231fd0fa46"
      },
      "source": [
        "y_pred = gs.predict(X_val_S)\n",
        "cm = confusion_matrix(y_val, y_pred)\n",
        "pd.DataFrame(cm, index=gs.best_estimator_.classes_, columns=gs.best_estimator_.classes_)"
      ],
      "execution_count": 156,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>0</th>\n",
              "      <th>1</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1536</td>\n",
              "      <td>51</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>162</td>\n",
              "      <td>1385</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      0     1\n",
              "0  1536    51\n",
              "1   162  1385"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 156
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "u7JbykEJz20z",
        "outputId": "cc760ebe-3c3a-4e53-c368-72a5abeb86d8"
      },
      "source": [
        "precision = precision_score(y_val, y_pred, pos_label=0)\n",
        "recall = recall_score(y_val, y_pred, pos_label=0)\n",
        "f1 = f1_score(y_val, y_pred, average=\"macro\")\n",
        "\n",
        "print(f\"precision: {precision}, recall: {recall}, f1-score: {f1}\")"
      ],
      "execution_count": 157,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "precision: 0.9045936395759717, recall: 0.9678638941398866, f1-score: 0.9318775955681893\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Ig94v053-7Dy",
        "outputId": "ddbe006a-e6ba-4783-faa3-f372d518491d"
      },
      "source": [
        "i = accuracy_interval(f1)\n",
        "accuracy[\"Support Vector Machines\"] = i\n",
        "print(i)"
      ],
      "execution_count": 158,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(0.9225, 0.9402)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "St5uh9Poz8q5",
        "outputId": "def3aa35-cfff-4ec7-e922-88695653bd78"
      },
      "source": [
        "svm_mse = mean_squared_error(y_val, y_pred)\n",
        "print('MSE: {}'.format(svm_mse))"
      ],
      "execution_count": 159,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "MSE: 0.06796426292278239\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rtRTwYxcTJoq"
      },
      "source": [
        "##Decision tree"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "O3Jd39oH_CrW"
      },
      "source": [
        "I parametri che andiamo a testare nella grid search sono:\n",
        "- `min_samples_split` che è il numero minimo di campioni che deve avere una foglia per poter essere splittata\n",
        "- `min_samples_leaf` numero minimo di campioni per ogni foglia (i.e. se splittiamo un nodo questo split è valido solo se lascia in ogni foglia che crea almeno min_samples_leaf campioni)\n",
        "- `max_depth` profondità massima che può raggiungere l'albero (con `None` cresce senza limiti su questo parametro)\n",
        "- `max_features` il numero massimo di feature da considerare per ogni split"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "9WtaBXAuRvK7",
        "outputId": "df6132b3-3c3d-4ad4-982a-819259eaba0b"
      },
      "source": [
        "model = DecisionTreeClassifier(random_state=42)\n",
        "\n",
        "num_features = X_train_S.shape[1]\n",
        "\n",
        "parameters = {\n",
        "             'min_samples_split': range(2, 4, 1),\n",
        "             'min_samples_leaf': range(1, 4, 1), \n",
        "             'max_depth': [None] + [i for i in range(2, 7)],\n",
        "             'max_features': range(2, num_features, 1)}\n",
        "\n",
        "skf = StratifiedKFold(3, shuffle=True, random_state=42)\n",
        "gs = GridSearchCV(model, parameters, cv=skf)\n",
        "\n",
        "gs.fit(X_train_S, y_train)"
      ],
      "execution_count": 160,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "GridSearchCV(cv=StratifiedKFold(n_splits=3, random_state=42, shuffle=True),\n",
              "             error_score=nan,\n",
              "             estimator=DecisionTreeClassifier(ccp_alpha=0.0, class_weight=None,\n",
              "                                              criterion='gini', max_depth=None,\n",
              "                                              max_features=None,\n",
              "                                              max_leaf_nodes=None,\n",
              "                                              min_impurity_decrease=0.0,\n",
              "                                              min_impurity_split=None,\n",
              "                                              min_samples_leaf=1,\n",
              "                                              min_samples_split=2,\n",
              "                                              min_weight_fraction_leaf=0.0,\n",
              "                                              presort='deprecated',\n",
              "                                              random_state=42,\n",
              "                                              splitter='best'),\n",
              "             iid='deprecated', n_jobs=None,\n",
              "             param_grid={'max_depth': [None, 2, 3, 4, 5, 6],\n",
              "                         'max_features': range(2, 20),\n",
              "                         'min_samples_leaf': range(1, 4),\n",
              "                         'min_samples_split': range(2, 4)},\n",
              "             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
              "             scoring=None, verbose=0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 160
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-fNTHi7VSVm0",
        "outputId": "bcfa0a54-7219-473c-c671-82faa1c8e396"
      },
      "source": [
        "gs.best_score_"
      ],
      "execution_count": 161,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.9521222581092837"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 161
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 111
        },
        "id": "U3Hx35c6Sdeh",
        "outputId": "7ece623a-9860-4a1a-e35f-1072982034ed"
      },
      "source": [
        "y_pred = gs.predict(X_val_S)\n",
        "cm = confusion_matrix(y_val, y_pred)\n",
        "pd.DataFrame(cm, index=gs.best_estimator_.classes_, columns=gs.best_estimator_.classes_)"
      ],
      "execution_count": 162,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>0</th>\n",
              "      <th>1</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1532</td>\n",
              "      <td>55</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>110</td>\n",
              "      <td>1437</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      0     1\n",
              "0  1532    55\n",
              "1   110  1437"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 162
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_Mbp3TkXSiPX",
        "outputId": "78ada9aa-a8f5-4c5a-8d0a-b776c7b45f17"
      },
      "source": [
        "precision = precision_score(y_val, y_pred, pos_label=0)\n",
        "recall = recall_score(y_val, y_pred, pos_label=0)\n",
        "f1 = f1_score(y_val, y_pred, average=\"macro\")\n",
        "\n",
        "print(f\"precision: {precision}, recall: {recall}, f1-score: {f1}\")"
      ],
      "execution_count": 163,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "precision: 0.9330085261875761, recall: 0.9653434152488973, f1-score: 0.9473032063508854\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vwEfcj_4SogG",
        "outputId": "870baa54-4cf4-40cb-f97f-9e7a0b1d0d47"
      },
      "source": [
        "i = accuracy_interval(f1)\n",
        "accuracy[\"Decision tree\"] = i\n",
        "print(i)"
      ],
      "execution_count": 164,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(0.9389, 0.9546)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qVWgGqTeStpi",
        "outputId": "1f9d4464-5c2b-43b1-a5e9-1a3228071b51"
      },
      "source": [
        "decisiontree_mse = mean_squared_error(y_val, y_pred)\n",
        "print('MSE: {}'.format(decisiontree_mse))"
      ],
      "execution_count": 165,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "MSE: 0.05264837268666241\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XfXdZVnlKvXI"
      },
      "source": [
        "###Visualizzazione dell'albero decisionale\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-LDwDn1tMC1q"
      },
      "source": [
        "Nei nodi troviamo:\n",
        "\n",
        "- Il criterio con cui viene effettuato il taglio \n",
        "\n",
        "\n",
        "- Il parametro `gini` che indica la qualità della suddivisione, rappresenta la frequenza con cui un elemento scelto casualmente dall'insieme verrebbe etichettato in modo errato se fosse etichettato casualmente in base alla distribuzione delle etichette nel sottoinsieme, è calcolato come:\n",
        "$$Gini = 1 - \\sum_{i=1}^C p_i^2 $$\n",
        "\n",
        "  $\\Rightarrow$ Assume valore 0 quando tutte le istanze nel nodo hanno una stessa label\n",
        "\n",
        "- Il parametro `samples` che indica la percentuale di campioni presenti in quel nodo\n",
        "- Il parametro `value` che indica la percentuale di istanze per ogni classe"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 654
        },
        "id": "T18hJT2UJb7K",
        "outputId": "c088b99f-c4ac-4879-ac0e-bb763ce60025"
      },
      "source": [
        "dot_data = tree.export_graphviz(gs.best_estimator_, out_file=None, \n",
        "                                feature_names=X.columns,  \n",
        "                                filled=True,\n",
        "                                proportion=True,\n",
        "                                max_depth=3)\n",
        "\n",
        "# Draw graph\n",
        "graph = graphviz.Source(dot_data, format=\"png\") \n",
        "graph"
      ],
      "execution_count": 166,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<graphviz.files.Source at 0x7fe880f96e90>"
            ],
            "image/svg+xml": "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"no\"?>\n<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n<!-- Generated by graphviz version 2.40.1 (20161225.0304)\n -->\n<!-- Title: Tree Pages: 1 -->\n<svg width=\"1713pt\" height=\"460pt\"\n viewBox=\"0.00 0.00 1712.50 460.00\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n<g id=\"graph0\" class=\"graph\" transform=\"scale(1 1) rotate(0) translate(4 456)\">\n<title>Tree</title>\n<polygon fill=\"#ffffff\" stroke=\"transparent\" points=\"-4,4 -4,-456 1708.5,-456 1708.5,4 -4,4\"/>\n<!-- 0 -->\n<g id=\"node1\" class=\"node\">\n<title>0</title>\n<polygon fill=\"#fcfeff\" stroke=\"#000000\" points=\"1000,-452 853,-452 853,-384 1000,-384 1000,-452\"/>\n<text text-anchor=\"middle\" x=\"926.5\" y=\"-436.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">age &lt;= &#45;0.273</text>\n<text text-anchor=\"middle\" x=\"926.5\" y=\"-421.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.5</text>\n<text text-anchor=\"middle\" x=\"926.5\" y=\"-406.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 100.0%</text>\n<text text-anchor=\"middle\" x=\"926.5\" y=\"-391.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.497, 0.503]</text>\n</g>\n<!-- 1 -->\n<g id=\"node2\" class=\"node\">\n<title>1</title>\n<polygon fill=\"#e88e4d\" stroke=\"#000000\" points=\"789,-348 642,-348 642,-280 789,-280 789,-348\"/>\n<text text-anchor=\"middle\" x=\"715.5\" y=\"-332.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">age &lt;= &#45;0.765</text>\n<text text-anchor=\"middle\" x=\"715.5\" y=\"-317.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.167</text>\n<text text-anchor=\"middle\" x=\"715.5\" y=\"-302.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 32.6%</text>\n<text text-anchor=\"middle\" x=\"715.5\" y=\"-287.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.908, 0.092]</text>\n</g>\n<!-- 0&#45;&gt;1 -->\n<g id=\"edge1\" class=\"edge\">\n<title>0&#45;&gt;1</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M857.4107,-383.9465C837.0998,-373.9354 814.7367,-362.9129 793.8888,-352.6371\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"795.2615,-349.4117 784.7445,-348.13 792.1668,-355.6905 795.2615,-349.4117\"/>\n<text text-anchor=\"middle\" x=\"792.7853\" y=\"-368.1016\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">True</text>\n</g>\n<!-- 112 -->\n<g id=\"node17\" class=\"node\">\n<title>112</title>\n<polygon fill=\"#8dc7f0\" stroke=\"#000000\" points=\"1186,-348 1033,-348 1033,-280 1186,-280 1186,-348\"/>\n<text text-anchor=\"middle\" x=\"1109.5\" y=\"-332.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">hypertension &lt;= &#45;0.539</text>\n<text text-anchor=\"middle\" x=\"1109.5\" y=\"-317.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.418</text>\n<text text-anchor=\"middle\" x=\"1109.5\" y=\"-302.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 67.4%</text>\n<text text-anchor=\"middle\" x=\"1109.5\" y=\"-287.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.298, 0.702]</text>\n</g>\n<!-- 0&#45;&gt;112 -->\n<g id=\"edge16\" class=\"edge\">\n<title>0&#45;&gt;112</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M986.421,-383.9465C1003.6417,-374.1599 1022.5633,-363.4066 1040.2954,-353.3294\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1042.4796,-356.1139 1049.4443,-348.13 1039.0209,-350.028 1042.4796,-356.1139\"/>\n<text text-anchor=\"middle\" x=\"1042.8096\" y=\"-368.5335\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">False</text>\n</g>\n<!-- 2 -->\n<g id=\"node3\" class=\"node\">\n<title>2</title>\n<polygon fill=\"#e5833c\" stroke=\"#000000\" points=\"467.5,-244 249.5,-244 249.5,-176 467.5,-176 467.5,-244\"/>\n<text text-anchor=\"middle\" x=\"358.5\" y=\"-228.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">smoking_status_smokes &lt;= &#45;0.473</text>\n<text text-anchor=\"middle\" x=\"358.5\" y=\"-213.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.033</text>\n<text text-anchor=\"middle\" x=\"358.5\" y=\"-198.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 21.9%</text>\n<text text-anchor=\"middle\" x=\"358.5\" y=\"-183.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.983, 0.017]</text>\n</g>\n<!-- 1&#45;&gt;2 -->\n<g id=\"edge2\" class=\"edge\">\n<title>1&#45;&gt;2</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M641.9622,-292.5772C594.4977,-278.75 531.6762,-260.4491 477.2381,-244.5904\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"478.1082,-241.1984 467.5283,-241.7618 476.1503,-247.919 478.1082,-241.1984\"/>\n</g>\n<!-- 21 -->\n<g id=\"node10\" class=\"node\">\n<title>21</title>\n<polygon fill=\"#edaa79\" stroke=\"#000000\" points=\"817,-244 614,-244 614,-176 817,-176 817,-244\"/>\n<text text-anchor=\"middle\" x=\"715.5\" y=\"-228.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">Residence_type_Rural &lt;= 1.104</text>\n<text text-anchor=\"middle\" x=\"715.5\" y=\"-213.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.37</text>\n<text text-anchor=\"middle\" x=\"715.5\" y=\"-198.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 10.8%</text>\n<text text-anchor=\"middle\" x=\"715.5\" y=\"-183.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.755, 0.245]</text>\n</g>\n<!-- 1&#45;&gt;21 -->\n<g id=\"edge9\" class=\"edge\">\n<title>1&#45;&gt;21</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M715.5,-279.9465C715.5,-271.776 715.5,-262.9318 715.5,-254.3697\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"719.0001,-254.13 715.5,-244.13 712.0001,-254.13 719.0001,-254.13\"/>\n</g>\n<!-- 3 -->\n<g id=\"node4\" class=\"node\">\n<title>3</title>\n<polygon fill=\"#e5813a\" stroke=\"#000000\" points=\"233,-140 0,-140 0,-72 233,-72 233,-140\"/>\n<text text-anchor=\"middle\" x=\"116.5\" y=\"-124.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">work_type_Self&#45;employed &lt;= &#45;0.313</text>\n<text text-anchor=\"middle\" x=\"116.5\" y=\"-109.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.007</text>\n<text text-anchor=\"middle\" x=\"116.5\" y=\"-94.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 19.1%</text>\n<text text-anchor=\"middle\" x=\"116.5\" y=\"-79.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.997, 0.003]</text>\n</g>\n<!-- 2&#45;&gt;3 -->\n<g id=\"edge3\" class=\"edge\">\n<title>2&#45;&gt;3</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M279.2601,-175.9465C255.5473,-165.7559 229.3956,-154.5171 205.1206,-144.0849\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"206.4873,-140.8628 195.9179,-140.13 203.7235,-147.294 206.4873,-140.8628\"/>\n</g>\n<!-- 18 -->\n<g id=\"node7\" class=\"node\">\n<title>18</title>\n<polygon fill=\"#e89151\" stroke=\"#000000\" points=\"465.5,-140 251.5,-140 251.5,-72 465.5,-72 465.5,-140\"/>\n<text text-anchor=\"middle\" x=\"358.5\" y=\"-124.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">smoking_status_smokes &lt;= 2.436</text>\n<text text-anchor=\"middle\" x=\"358.5\" y=\"-109.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.196</text>\n<text text-anchor=\"middle\" x=\"358.5\" y=\"-94.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 2.8%</text>\n<text text-anchor=\"middle\" x=\"358.5\" y=\"-79.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.89, 0.11]</text>\n</g>\n<!-- 2&#45;&gt;18 -->\n<g id=\"edge6\" class=\"edge\">\n<title>2&#45;&gt;18</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M358.5,-175.9465C358.5,-167.776 358.5,-158.9318 358.5,-150.3697\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"362.0001,-150.13 358.5,-140.13 355.0001,-150.13 362.0001,-150.13\"/>\n</g>\n<!-- 4 -->\n<g id=\"node5\" class=\"node\">\n<title>4</title>\n<polygon fill=\"#c0c0c0\" stroke=\"#000000\" points=\"107.5,-36 53.5,-36 53.5,0 107.5,0 107.5,-36\"/>\n<text text-anchor=\"middle\" x=\"80.5\" y=\"-14.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">(...)</text>\n</g>\n<!-- 3&#45;&gt;4 -->\n<g id=\"edge4\" class=\"edge\">\n<title>3&#45;&gt;4</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M102.5815,-71.9769C98.9977,-63.2167 95.1862,-53.8995 91.7564,-45.5157\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"94.968,-44.1222 87.9422,-36.192 88.4891,-46.7727 94.968,-44.1222\"/>\n</g>\n<!-- 13 -->\n<g id=\"node6\" class=\"node\">\n<title>13</title>\n<polygon fill=\"#c0c0c0\" stroke=\"#000000\" points=\"179.5,-36 125.5,-36 125.5,0 179.5,0 179.5,-36\"/>\n<text text-anchor=\"middle\" x=\"152.5\" y=\"-14.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">(...)</text>\n</g>\n<!-- 3&#45;&gt;13 -->\n<g id=\"edge5\" class=\"edge\">\n<title>3&#45;&gt;13</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M130.4185,-71.9769C134.0023,-63.2167 137.8138,-53.8995 141.2436,-45.5157\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"144.5109,-46.7727 145.0578,-36.192 138.032,-44.1222 144.5109,-46.7727\"/>\n</g>\n<!-- 19 -->\n<g id=\"node8\" class=\"node\">\n<title>19</title>\n<polygon fill=\"#c0c0c0\" stroke=\"#000000\" points=\"349.5,-36 295.5,-36 295.5,0 349.5,0 349.5,-36\"/>\n<text text-anchor=\"middle\" x=\"322.5\" y=\"-14.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">(...)</text>\n</g>\n<!-- 18&#45;&gt;19 -->\n<g id=\"edge7\" class=\"edge\">\n<title>18&#45;&gt;19</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M344.5815,-71.9769C340.9977,-63.2167 337.1862,-53.8995 333.7564,-45.5157\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"336.968,-44.1222 329.9422,-36.192 330.4891,-46.7727 336.968,-44.1222\"/>\n</g>\n<!-- 20 -->\n<g id=\"node9\" class=\"node\">\n<title>20</title>\n<polygon fill=\"#c0c0c0\" stroke=\"#000000\" points=\"421.5,-36 367.5,-36 367.5,0 421.5,0 421.5,-36\"/>\n<text text-anchor=\"middle\" x=\"394.5\" y=\"-14.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">(...)</text>\n</g>\n<!-- 18&#45;&gt;20 -->\n<g id=\"edge8\" class=\"edge\">\n<title>18&#45;&gt;20</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M372.4185,-71.9769C376.0023,-63.2167 379.8138,-53.8995 383.2436,-45.5157\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"386.5109,-46.7727 387.0578,-36.192 380.032,-44.1222 386.5109,-46.7727\"/>\n</g>\n<!-- 22 -->\n<g id=\"node11\" class=\"node\">\n<title>22</title>\n<polygon fill=\"#f4c8a9\" stroke=\"#000000\" points=\"739.5,-140 483.5,-140 483.5,-72 739.5,-72 739.5,-140\"/>\n<text text-anchor=\"middle\" x=\"611.5\" y=\"-124.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">smoking_status_never smoked &lt;= &#45;0.853</text>\n<text text-anchor=\"middle\" x=\"611.5\" y=\"-109.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.462</text>\n<text text-anchor=\"middle\" x=\"611.5\" y=\"-94.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 6.4%</text>\n<text text-anchor=\"middle\" x=\"611.5\" y=\"-79.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.638, 0.362]</text>\n</g>\n<!-- 21&#45;&gt;22 -->\n<g id=\"edge10\" class=\"edge\">\n<title>21&#45;&gt;22</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M681.4465,-175.9465C672.2884,-166.7884 662.2838,-156.7838 652.7788,-147.2788\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"655.176,-144.7262 645.63,-140.13 650.2262,-149.676 655.176,-144.7262\"/>\n</g>\n<!-- 73 -->\n<g id=\"node14\" class=\"node\">\n<title>73</title>\n<polygon fill=\"#e78b49\" stroke=\"#000000\" points=\"943,-140 758,-140 758,-72 943,-72 943,-140\"/>\n<text text-anchor=\"middle\" x=\"850.5\" y=\"-124.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">avg_glucose_level &lt;= &#45;0.975</text>\n<text text-anchor=\"middle\" x=\"850.5\" y=\"-109.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.141</text>\n<text text-anchor=\"middle\" x=\"850.5\" y=\"-94.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 4.4%</text>\n<text text-anchor=\"middle\" x=\"850.5\" y=\"-79.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.924, 0.076]</text>\n</g>\n<!-- 21&#45;&gt;73 -->\n<g id=\"edge13\" class=\"edge\">\n<title>21&#45;&gt;73</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M759.704,-175.9465C771.9416,-166.519 785.3436,-156.1946 798.0038,-146.4415\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"800.4107,-149.0055 806.1966,-140.13 796.1388,-143.4601 800.4107,-149.0055\"/>\n</g>\n<!-- 23 -->\n<g id=\"node12\" class=\"node\">\n<title>23</title>\n<polygon fill=\"#c0c0c0\" stroke=\"#000000\" points=\"602.5,-36 548.5,-36 548.5,0 602.5,0 602.5,-36\"/>\n<text text-anchor=\"middle\" x=\"575.5\" y=\"-14.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">(...)</text>\n</g>\n<!-- 22&#45;&gt;23 -->\n<g id=\"edge11\" class=\"edge\">\n<title>22&#45;&gt;23</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M597.5815,-71.9769C593.9977,-63.2167 590.1862,-53.8995 586.7564,-45.5157\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"589.968,-44.1222 582.9422,-36.192 583.4891,-46.7727 589.968,-44.1222\"/>\n</g>\n<!-- 42 -->\n<g id=\"node13\" class=\"node\">\n<title>42</title>\n<polygon fill=\"#c0c0c0\" stroke=\"#000000\" points=\"674.5,-36 620.5,-36 620.5,0 674.5,0 674.5,-36\"/>\n<text text-anchor=\"middle\" x=\"647.5\" y=\"-14.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">(...)</text>\n</g>\n<!-- 22&#45;&gt;42 -->\n<g id=\"edge12\" class=\"edge\">\n<title>22&#45;&gt;42</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M625.4185,-71.9769C629.0023,-63.2167 632.8138,-53.8995 636.2436,-45.5157\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"639.5109,-46.7727 640.0578,-36.192 633.032,-44.1222 639.5109,-46.7727\"/>\n</g>\n<!-- 74 -->\n<g id=\"node15\" class=\"node\">\n<title>74</title>\n<polygon fill=\"#c0c0c0\" stroke=\"#000000\" points=\"825.5,-36 771.5,-36 771.5,0 825.5,0 825.5,-36\"/>\n<text text-anchor=\"middle\" x=\"798.5\" y=\"-14.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">(...)</text>\n</g>\n<!-- 73&#45;&gt;74 -->\n<g id=\"edge14\" class=\"edge\">\n<title>73&#45;&gt;74</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M830.3954,-71.9769C825.1064,-63.0262 819.4738,-53.4941 814.437,-44.9703\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"817.3504,-43.0207 809.2498,-36.192 811.3239,-46.5818 817.3504,-43.0207\"/>\n</g>\n<!-- 85 -->\n<g id=\"node16\" class=\"node\">\n<title>85</title>\n<polygon fill=\"#c0c0c0\" stroke=\"#000000\" points=\"897.5,-36 843.5,-36 843.5,0 897.5,0 897.5,-36\"/>\n<text text-anchor=\"middle\" x=\"870.5\" y=\"-14.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">(...)</text>\n</g>\n<!-- 73&#45;&gt;85 -->\n<g id=\"edge15\" class=\"edge\">\n<title>73&#45;&gt;85</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M858.2325,-71.9769C860.1802,-63.4071 862.249,-54.3043 864.1219,-46.0638\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"867.5621,-46.719 866.3655,-36.192 860.7362,-45.1676 867.5621,-46.719\"/>\n</g>\n<!-- 113 -->\n<g id=\"node18\" class=\"node\">\n<title>113</title>\n<polygon fill=\"#c5e2f7\" stroke=\"#000000\" points=\"1203.5,-244 1015.5,-244 1015.5,-176 1203.5,-176 1203.5,-244\"/>\n<text text-anchor=\"middle\" x=\"1109.5\" y=\"-228.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">work_type_Private &lt;= &#45;1.292</text>\n<text text-anchor=\"middle\" x=\"1109.5\" y=\"-213.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.485</text>\n<text text-anchor=\"middle\" x=\"1109.5\" y=\"-198.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 40.7%</text>\n<text text-anchor=\"middle\" x=\"1109.5\" y=\"-183.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.414, 0.586]</text>\n</g>\n<!-- 112&#45;&gt;113 -->\n<g id=\"edge17\" class=\"edge\">\n<title>112&#45;&gt;113</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M1109.5,-279.9465C1109.5,-271.776 1109.5,-262.9318 1109.5,-254.3697\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1113.0001,-254.13 1109.5,-244.13 1106.0001,-254.13 1113.0001,-254.13\"/>\n</g>\n<!-- 462 -->\n<g id=\"node25\" class=\"node\">\n<title>462</title>\n<polygon fill=\"#54aae9\" stroke=\"#000000\" points=\"1446,-244 1297,-244 1297,-176 1446,-176 1446,-244\"/>\n<text text-anchor=\"middle\" x=\"1371.5\" y=\"-228.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">hypertension &lt;= 2.341</text>\n<text text-anchor=\"middle\" x=\"1371.5\" y=\"-213.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.212</text>\n<text text-anchor=\"middle\" x=\"1371.5\" y=\"-198.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 26.7%</text>\n<text text-anchor=\"middle\" x=\"1371.5\" y=\"-183.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.121, 0.879]</text>\n</g>\n<!-- 112&#45;&gt;462 -->\n<g id=\"edge24\" class=\"edge\">\n<title>112&#45;&gt;462</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M1186.3384,-283.4993C1218.0841,-270.8979 1254.9804,-256.2521 1287.4441,-243.3657\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1288.9158,-246.5472 1296.919,-239.6047 1286.3332,-240.0411 1288.9158,-246.5472\"/>\n</g>\n<!-- 114 -->\n<g id=\"node19\" class=\"node\">\n<title>114</title>\n<polygon fill=\"#f1bd98\" stroke=\"#000000\" points=\"1108,-140 961,-140 961,-72 1108,-72 1108,-140\"/>\n<text text-anchor=\"middle\" x=\"1034.5\" y=\"-124.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">age &lt;= 0.549</text>\n<text text-anchor=\"middle\" x=\"1034.5\" y=\"-109.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.438</text>\n<text text-anchor=\"middle\" x=\"1034.5\" y=\"-94.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 10.8%</text>\n<text text-anchor=\"middle\" x=\"1034.5\" y=\"-79.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.676, 0.324]</text>\n</g>\n<!-- 113&#45;&gt;114 -->\n<g id=\"edge18\" class=\"edge\">\n<title>113&#45;&gt;114</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M1084.9422,-175.9465C1078.5968,-167.1475 1071.6879,-157.5672 1065.0764,-148.3993\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1067.8011,-146.1937 1059.113,-140.13 1062.1234,-150.2882 1067.8011,-146.1937\"/>\n</g>\n<!-- 241 -->\n<g id=\"node22\" class=\"node\">\n<title>241</title>\n<polygon fill=\"#96cbf1\" stroke=\"#000000\" points=\"1295,-140 1126,-140 1126,-72 1295,-72 1295,-140\"/>\n<text text-anchor=\"middle\" x=\"1210.5\" y=\"-124.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">work_type_Private &lt;= 0.9</text>\n<text text-anchor=\"middle\" x=\"1210.5\" y=\"-109.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.435</text>\n<text text-anchor=\"middle\" x=\"1210.5\" y=\"-94.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 29.9%</text>\n<text text-anchor=\"middle\" x=\"1210.5\" y=\"-79.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.319, 0.681]</text>\n</g>\n<!-- 113&#45;&gt;241 -->\n<g id=\"edge21\" class=\"edge\">\n<title>113&#45;&gt;241</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M1142.5712,-175.9465C1151.3779,-166.8782 1160.9907,-156.9799 1170.1403,-147.5585\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1172.8985,-149.7422 1177.3545,-140.13 1167.8768,-144.8654 1172.8985,-149.7422\"/>\n</g>\n<!-- 115 -->\n<g id=\"node20\" class=\"node\">\n<title>115</title>\n<polygon fill=\"#c0c0c0\" stroke=\"#000000\" points=\"1005.5,-36 951.5,-36 951.5,0 1005.5,0 1005.5,-36\"/>\n<text text-anchor=\"middle\" x=\"978.5\" y=\"-14.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">(...)</text>\n</g>\n<!-- 114&#45;&gt;115 -->\n<g id=\"edge19\" class=\"edge\">\n<title>114&#45;&gt;115</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M1012.8489,-71.9769C1007.0925,-62.931 1000.9581,-53.2913 995.49,-44.6986\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"998.3983,-42.7495 990.0767,-36.192 992.4927,-46.5077 998.3983,-42.7495\"/>\n</g>\n<!-- 158 -->\n<g id=\"node21\" class=\"node\">\n<title>158</title>\n<polygon fill=\"#c0c0c0\" stroke=\"#000000\" points=\"1077.5,-36 1023.5,-36 1023.5,0 1077.5,0 1077.5,-36\"/>\n<text text-anchor=\"middle\" x=\"1050.5\" y=\"-14.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">(...)</text>\n</g>\n<!-- 114&#45;&gt;158 -->\n<g id=\"edge20\" class=\"edge\">\n<title>114&#45;&gt;158</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M1040.686,-71.9769C1042.2442,-63.4071 1043.8992,-54.3043 1045.3975,-46.0638\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1048.847,-46.6568 1047.1924,-36.192 1041.9599,-45.4045 1048.847,-46.6568\"/>\n</g>\n<!-- 242 -->\n<g id=\"node23\" class=\"node\">\n<title>242</title>\n<polygon fill=\"#c0c0c0\" stroke=\"#000000\" points=\"1201.5,-36 1147.5,-36 1147.5,0 1201.5,0 1201.5,-36\"/>\n<text text-anchor=\"middle\" x=\"1174.5\" y=\"-14.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">(...)</text>\n</g>\n<!-- 241&#45;&gt;242 -->\n<g id=\"edge22\" class=\"edge\">\n<title>241&#45;&gt;242</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M1196.5815,-71.9769C1192.9977,-63.2167 1189.1862,-53.8995 1185.7564,-45.5157\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1188.968,-44.1222 1181.9422,-36.192 1182.4891,-46.7727 1188.968,-44.1222\"/>\n</g>\n<!-- 243 -->\n<g id=\"node24\" class=\"node\">\n<title>243</title>\n<polygon fill=\"#c0c0c0\" stroke=\"#000000\" points=\"1273.5,-36 1219.5,-36 1219.5,0 1273.5,0 1273.5,-36\"/>\n<text text-anchor=\"middle\" x=\"1246.5\" y=\"-14.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">(...)</text>\n</g>\n<!-- 241&#45;&gt;243 -->\n<g id=\"edge23\" class=\"edge\">\n<title>241&#45;&gt;243</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M1224.4185,-71.9769C1228.0023,-63.2167 1231.8138,-53.8995 1235.2436,-45.5157\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1238.5109,-46.7727 1239.0578,-36.192 1232.032,-44.1222 1238.5109,-46.7727\"/>\n</g>\n<!-- 463 -->\n<g id=\"node26\" class=\"node\">\n<title>463</title>\n<polygon fill=\"#399de5\" stroke=\"#000000\" points=\"1430,-132.5 1313,-132.5 1313,-79.5 1430,-79.5 1430,-132.5\"/>\n<text text-anchor=\"middle\" x=\"1371.5\" y=\"-117.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.0</text>\n<text text-anchor=\"middle\" x=\"1371.5\" y=\"-102.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 18.4%</text>\n<text text-anchor=\"middle\" x=\"1371.5\" y=\"-87.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.0, 1.0]</text>\n</g>\n<!-- 462&#45;&gt;463 -->\n<g id=\"edge25\" class=\"edge\">\n<title>462&#45;&gt;463</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M1371.5,-175.9465C1371.5,-165.2621 1371.5,-153.4254 1371.5,-142.5742\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1375.0001,-142.5421 1371.5,-132.5422 1368.0001,-142.5422 1375.0001,-142.5421\"/>\n</g>\n<!-- 464 -->\n<g id=\"node27\" class=\"node\">\n<title>464</title>\n<polygon fill=\"#b6dbf5\" stroke=\"#000000\" points=\"1704.5,-140 1448.5,-140 1448.5,-72 1704.5,-72 1704.5,-140\"/>\n<text text-anchor=\"middle\" x=\"1576.5\" y=\"-124.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">smoking_status_never smoked &lt;= &#45;0.857</text>\n<text text-anchor=\"middle\" x=\"1576.5\" y=\"-109.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">gini = 0.474</text>\n<text text-anchor=\"middle\" x=\"1576.5\" y=\"-94.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">samples = 8.3%</text>\n<text text-anchor=\"middle\" x=\"1576.5\" y=\"-79.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">value = [0.387, 0.613]</text>\n</g>\n<!-- 462&#45;&gt;464 -->\n<g id=\"edge26\" class=\"edge\">\n<title>462&#45;&gt;464</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M1438.6247,-175.9465C1458.2695,-165.9803 1479.8904,-155.0117 1500.0677,-144.7754\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1501.89,-147.7756 1509.2245,-140.13 1498.723,-141.533 1501.89,-147.7756\"/>\n</g>\n<!-- 465 -->\n<g id=\"node28\" class=\"node\">\n<title>465</title>\n<polygon fill=\"#c0c0c0\" stroke=\"#000000\" points=\"1567.5,-36 1513.5,-36 1513.5,0 1567.5,0 1567.5,-36\"/>\n<text text-anchor=\"middle\" x=\"1540.5\" y=\"-14.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">(...)</text>\n</g>\n<!-- 464&#45;&gt;465 -->\n<g id=\"edge27\" class=\"edge\">\n<title>464&#45;&gt;465</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M1562.5815,-71.9769C1558.9977,-63.2167 1555.1862,-53.8995 1551.7564,-45.5157\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1554.968,-44.1222 1547.9422,-36.192 1548.4891,-46.7727 1554.968,-44.1222\"/>\n</g>\n<!-- 514 -->\n<g id=\"node29\" class=\"node\">\n<title>514</title>\n<polygon fill=\"#c0c0c0\" stroke=\"#000000\" points=\"1639.5,-36 1585.5,-36 1585.5,0 1639.5,0 1639.5,-36\"/>\n<text text-anchor=\"middle\" x=\"1612.5\" y=\"-14.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">(...)</text>\n</g>\n<!-- 464&#45;&gt;514 -->\n<g id=\"edge28\" class=\"edge\">\n<title>464&#45;&gt;514</title>\n<path fill=\"none\" stroke=\"#000000\" d=\"M1590.4185,-71.9769C1594.0023,-63.2167 1597.8138,-53.8995 1601.2436,-45.5157\"/>\n<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1604.5109,-46.7727 1605.0578,-36.192 1598.032,-44.1222 1604.5109,-46.7727\"/>\n</g>\n</g>\n</svg>\n"
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 166
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5I8VeragTjuM"
      },
      "source": [
        "##XGBoost"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "6birO_a1TiaK",
        "outputId": "cab98f5d-da86-4b95-d884-b0dcf245538c"
      },
      "source": [
        "model = XGBClassifier(nthread=8, objective='binary:logistic')\n",
        "\n",
        "\n",
        "parameters = {\n",
        "    'eta': [0.002, 0.1, 0.5],\n",
        "    'max_depth': [6],\n",
        "    'n_estimators': [150, 300],\n",
        "    'alpha': [0.0001, 0.001]\n",
        "}\n",
        "\n",
        "skf = StratifiedKFold(3, shuffle=True, random_state=42)\n",
        "gs = GridSearchCV(model, parameters, cv=skf)\n",
        "\n",
        "gs.fit(X_train_S, y_train)"
      ],
      "execution_count": 167,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "GridSearchCV(cv=StratifiedKFold(n_splits=3, random_state=42, shuffle=True),\n",
              "             error_score=nan,\n",
              "             estimator=XGBClassifier(base_score=0.5, booster='gbtree',\n",
              "                                     colsample_bylevel=1, colsample_bynode=1,\n",
              "                                     colsample_bytree=1, gamma=0,\n",
              "                                     learning_rate=0.1, max_delta_step=0,\n",
              "                                     max_depth=3, min_child_weight=1,\n",
              "                                     missing=None, n_estimators=100, n_jobs=1,\n",
              "                                     nthread=8, objective='binary:logistic',\n",
              "                                     random_state=0, reg_alpha=0, reg_lambda=1,\n",
              "                                     scale_pos_weight=1, seed=None, silent=None,\n",
              "                                     subsample=1, verbosity=1),\n",
              "             iid='deprecated', n_jobs=None,\n",
              "             param_grid={'alpha': [0.0001, 0.001], 'eta': [0.002, 0.1, 0.5],\n",
              "                         'max_depth': [6], 'n_estimators': [150, 300]},\n",
              "             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
              "             scoring=None, verbose=0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 167
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qSlWzn7wUCwC",
        "outputId": "990ea04b-8e37-4d44-be84-3c7140793899"
      },
      "source": [
        "gs.best_score_"
      ],
      "execution_count": 168,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.9720719031208294"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 168
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 111
        },
        "id": "EiAS-agXUEc1",
        "outputId": "1f695ac7-c705-4378-f698-65ff7eb2f143"
      },
      "source": [
        "y_pred = gs.predict(X_val_S)\n",
        "cm = confusion_matrix(y_val, y_pred)\n",
        "pd.DataFrame(cm, index=gs.best_estimator_.classes_, columns=gs.best_estimator_.classes_)"
      ],
      "execution_count": 169,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>0</th>\n",
              "      <th>1</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1579</td>\n",
              "      <td>8</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>78</td>\n",
              "      <td>1469</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      0     1\n",
              "0  1579     8\n",
              "1    78  1469"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 169
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ITw95R8vUIJa",
        "outputId": "56edb654-b2b0-4c69-acb5-c76e193f8145"
      },
      "source": [
        "precision = precision_score(y_val, y_pred, pos_label=0)\n",
        "recall = recall_score(y_val, y_pred, pos_label=0)\n",
        "f1 = f1_score(y_val, y_pred, average=\"macro\")\n",
        "\n",
        "print(f\"precision: {precision}, recall: {recall}, f1-score: {f1}\")"
      ],
      "execution_count": 170,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "precision: 0.9529269764634882, recall: 0.9949590422180214, f1-score: 0.9725251828365269\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ryhbcOEpULrN",
        "outputId": "c34e7f47-fd5d-44ac-d7cb-d442f1420b0c"
      },
      "source": [
        "i = accuracy_interval(f1)\n",
        "accuracy[\"XGBoost\"] = i\n",
        "print(i)"
      ],
      "execution_count": 171,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(0.9662, 0.9777)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vBNe-pKkUP5j",
        "outputId": "70baadc4-f8be-4baf-9945-96255873a654"
      },
      "source": [
        "xgboost_mse = mean_squared_error(y_val, y_pred)\n",
        "print('MSE: {}'.format(decisiontree_mse))"
      ],
      "execution_count": 172,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "MSE: 0.05264837268666241\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "n9aGBM1ebfyy"
      },
      "source": [
        "##Model comparison"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gOMM8jHbEqPR"
      },
      "source": [
        "Prendiamo una confidenza del 95%, di conseguenza dalla tabella della distribuzione z otteniamo 1.96 come valore per `z` "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "y9b8FBVIFVg4"
      },
      "source": [
        "La seguente funzione implementa il confronto fra due modelli dati i rispettivi errori `e1` ed `e2` e la cardinalità del test set `n` secondo le formule :\n",
        "\n",
        "$$ d = |e_1 - e_2| $$\n",
        "\n",
        "$$ \\sigma_t^2 = \\sigma_1^2 + \\sigma_2^2 = \\frac{e_1(1-e_1)}{n}  + \\frac{e_2(1-e_2)}{n}$$\n",
        "\n",
        "$$d_t = d \\pm z_{1-\\alpha} \\cdot \\sigma_t$$"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cXf-LLmX0veh"
      },
      "source": [
        "def intervallo(mse1, mse2):\n",
        "  d = np.abs(mse1 - mse2)\n",
        "  variance = (mse1 * (1 - mse1)) / len(X_val) + (mse2 * (1 - mse2)) / len(X_val)\n",
        "  d_min = d - 1.96 * np.sqrt(variance)\n",
        "  d_max = d + 1.96 * np.sqrt(variance)\n",
        "  return d_min, d_max"
      ],
      "execution_count": 173,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kQXPlulfmn20"
      },
      "source": [
        "Andiamo a calcolare l'intervallo fra tutte le coppie di modelli"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_VuoPBEljaWT",
        "outputId": "d69165a2-da2e-4d0d-d9dc-5ec029b019a2"
      },
      "source": [
        "from itertools import combinations\n",
        "mse = [(\"perceptron_mse\", \"Perceptron\"), (\"logisticregression_mse\",\"Logistic Regression\"),\n",
        "       (\"svm_mse\", \"SVM\"), (\"decisiontree_mse\", \"Decision Tree\"), (\"xgboost_mse\", \"XGBoost\")]\n",
        "\n",
        "print (f\"{'Models':<40} {'Interval':<15}\")\n",
        "for m1, m2 in list(combinations(mse, 2)):\n",
        "  mse1, mse2 = eval(m1[0]), eval(m2[0])\n",
        "  name1, name2 = m1[1], m2[1]\n",
        "  comparison = name1 + \" vs \" + name2\n",
        "  print (f\"{comparison:<40} {np.round(intervallo(mse1 , mse2), 4)} \")"
      ],
      "execution_count": 174,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Models                                   Interval       \n",
            "Perceptron vs Logistic Regression        [0.0599 0.1029] \n",
            "Perceptron vs SVM                        [0.2096 0.2461] \n",
            "Perceptron vs Decision Tree              [0.2254 0.2609] \n",
            "Perceptron vs XGBoost                    [0.2514 0.2853] \n",
            "Logistic Regression vs SVM               [0.1296 0.1633] \n",
            "Logistic Regression vs Decision Tree     [0.1454 0.1781] \n",
            "Logistic Regression vs XGBoost           [0.1715 0.2024] \n",
            "SVM vs Decision Tree                     [0.0035 0.0271] \n",
            "SVM vs XGBoost                           [0.03  0.051] \n",
            "Decision Tree vs XGBoost                 [0.0155 0.0349] \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tafqgBIUwbLg"
      },
      "source": [
        "###Confronto con un modello casuale"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dD0nSQw-oRNM"
      },
      "source": [
        "La seguente funzione calcola, come la precedente, l'intervallo per il confronto fra due modelli ma utilizza una confidenza al 99% invece del 95%"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fi1LaALw3pF6"
      },
      "source": [
        "def intervallo99(mse1, mse2):\n",
        "  d = np.abs(mse1 - mse2)\n",
        "  variance = (mse1 * (1 - mse1)) / len(X_val) + (mse2 * (1 - mse2)) / len(X_val)\n",
        "  d_min = d - 2.58 * np.sqrt(variance)\n",
        "  d_max = d + 2.58 * np.sqrt(variance)\n",
        "  return d_min, d_max"
      ],
      "execution_count": 175,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XK5eO8EgwdWy",
        "outputId": "e564d213-aa70-41cb-cdb9-b6285677df86"
      },
      "source": [
        "random = DummyClassifier(strategy=\"uniform\", random_state=42)\n",
        "random.fit(X_train_S, y_train)\n",
        "y_pred = random.predict(X_val_S)\n",
        "print(random.score(X_val_S, y_val))"
      ],
      "execution_count": 176,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.5028717294192725\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "io16GmDOwv8v",
        "outputId": "1d162873-585e-47fe-af82-695e1b542264"
      },
      "source": [
        "mse_random = mean_squared_error(y_val, y_pred)\n",
        "print(mse_random)"
      ],
      "execution_count": 177,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.4971282705807275\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wnWpv2DGnmhC",
        "outputId": "5b517f21-b5a3-4faf-85f5-53473ac05423"
      },
      "source": [
        "print (f\"{'Models':<40} Interval\")\n",
        "for m in mse:\n",
        "  mse_i = eval(m[0])\n",
        "  name_i = m[1]\n",
        "  comparison = name_i + \" vs Random\"\n",
        "  print (f\"{comparison:<40} {np.round(intervallo99(mse_i , mse_random), 4)} \")"
      ],
      "execution_count": 178,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Models                                   Interval\n",
            "Perceptron vs Random                     [0.1701 0.2325] \n",
            "Logistic Regression vs Random            [0.2529 0.3125] \n",
            "SVM vs Random                            [0.4034 0.455 ] \n",
            "Decision Tree vs Random                  [0.4192 0.4697] \n",
            "XGBoost vs Random                        [0.4454 0.4939] \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WkN9YQ-c3giu"
      },
      "source": [
        "Possiamo vedere come la differenza fra tutti i modelli e uno random sia sempre statisticamente significativa quindi i modelli implementati sono tutti accettabili "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "O3WPwUOmIejk"
      },
      "source": [
        "##Conclusioni"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5c03LctwIrGh"
      },
      "source": [
        "Dalla precedente sezione *Model Comparison* possiamo vedere che fra tutti i modelli la differenza è stasticamente significativa."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fyrF9rCQIsVr"
      },
      "source": [
        "Riassunto dei vari f1-score:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Jtil616UKPFu",
        "outputId": "43d662aa-a309-455a-cdce-d206a73424f7"
      },
      "source": [
        "print (f\"{'Model':<40} f-1 score\")\n",
        "for k in accuracy.keys():\n",
        "    print(f\"{k:<40} {accuracy[k]}\")"
      ],
      "execution_count": 179,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model                                    f-1 score\n",
            "Perceptron                               (0.7272, 0.7578)\n",
            "Perceptron with gs                       (0.6824, 0.7145)\n",
            "Logistic Regression                      (0.7708, 0.7996)\n",
            "Logistic regression with gs              (0.7708, 0.7996)\n",
            "Support Vector Machines                  (0.9225, 0.9402)\n",
            "Decision tree                            (0.9389, 0.9546)\n",
            "XGBoost                                  (0.9662, 0.9777)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ukfx4T0eqEsU"
      },
      "source": [
        "Visti i risultati ottenuti per l'indice *f1-score* per i vari modelli si può dire che i migliori sono:\n",
        "- SVM con kernel RBF non lineare\n",
        "- Decision tree\n",
        "- XGBoost\n",
        "\n",
        "Questo probabilmente è dovuto al fatto che inizialmente la correlazione fra le 11 feature di input è abbastanza bassa quindi usando modelli non lineari andiamo a separare meglio i dati.\n",
        "\n",
        "Questo lo si può notare anche dal fatto che se implementiamo un modello SVM con kernel lineare otteniamo un f-1 score molto più basso (come si vede nell'esempio di seguito) "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "PlbsOITDHPcJ",
        "outputId": "ec0a6f5c-8765-4b8b-8eee-1c0fe74d6bf8"
      },
      "source": [
        "model = SVC()\n",
        "parameters = {\n",
        "    \"kernel\": [\"linear\"],\n",
        "    \"C\": np.logspace(-2, 0, 3)\n",
        "}\n",
        "\n",
        "skf = StratifiedKFold(3, shuffle=True, random_state=42)\n",
        "gs = GridSearchCV(model, parameters, cv=skf)\n",
        "\n",
        "gs.fit(X_train_S, y_train)"
      ],
      "execution_count": 180,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "GridSearchCV(cv=StratifiedKFold(n_splits=3, random_state=42, shuffle=True),\n",
              "             error_score=nan,\n",
              "             estimator=SVC(C=1.0, break_ties=False, cache_size=200,\n",
              "                           class_weight=None, coef0=0.0,\n",
              "                           decision_function_shape='ovr', degree=3,\n",
              "                           gamma='scale', kernel='rbf', max_iter=-1,\n",
              "                           probability=False, random_state=None, shrinking=True,\n",
              "                           tol=0.001, verbose=False),\n",
              "             iid='deprecated', n_jobs=None,\n",
              "             param_grid={'C': array([0.01, 0.1 , 1.  ]), 'kernel': ['linear']},\n",
              "             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
              "             scoring=None, verbose=0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 180
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 111
        },
        "id": "cH9ohIAvHXTh",
        "outputId": "6772ac83-afd8-4ae7-fe44-b0545d1cec31"
      },
      "source": [
        "y_pred = gs.predict(X_val_S)\n",
        "cm = confusion_matrix(y_val, y_pred)\n",
        "pd.DataFrame(cm, index=gs.best_estimator_.classes_, columns=gs.best_estimator_.classes_)"
      ],
      "execution_count": 181,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>0</th>\n",
              "      <th>1</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1183</td>\n",
              "      <td>404</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>272</td>\n",
              "      <td>1275</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      0     1\n",
              "0  1183   404\n",
              "1   272  1275"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 181
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_a0IB1c-VvMA",
        "outputId": "cbeaf6bb-71a6-40f6-f80a-f00e3397a7cd"
      },
      "source": [
        "X_val.shape"
      ],
      "execution_count": 182,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(3134, 20)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 182
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gSpvmKnRHYzz",
        "outputId": "6af573df-f34c-4717-d29b-2b09a73cc0af"
      },
      "source": [
        "precision = precision_score(y_val, y_pred, pos_label=0)\n",
        "recall = recall_score(y_val, y_pred, pos_label=0)\n",
        "f1 = f1_score(y_val, y_pred, average=\"macro\")\n",
        "print(f\"precision: {precision}, recall: {recall}, f1-score: {f1}\")"
      ],
      "execution_count": 183,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "precision: 0.8130584192439863, recall: 0.7454316320100819, f1-score: 0.7841151753117035\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}