Scikit_learn.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "toc_visible": true,
      "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/vn322/Synergy_DBM_301_CS/blob/main/Scikit_learn.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Scikit-learn - одна из наиболее популярных библиотек Python для Data Science и Machine Learning. Она позволяет выполнять большое количество операций и предоставляет множество алгоритмов. Scikit-learn также предлагает отличную документацию, в том числе и на русском языке описывающую классы, методы и функции библиотеки, а также описание  алгоритмов с необходимыми сведениями из математики. \n",
        "\n",
        "Scikit-Learn имеет в своём составе методы для:\n",
        "\n",
        "- предварительной обработки данных;\n",
        "- уменьшения размерности;\n",
        "- подбора гиперпараметров  модели;\n",
        "- регрессии;\n",
        "- классификации;\n",
        "- кластерного анализа\n",
        "- и многое другое"
      ],
      "metadata": {
        "id": "yPtQG4h1bsrL"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns"
      ],
      "metadata": {
        "id": "j_yl9mwBeFcc"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Для реализации алгоритмов машинного обучения желательно иметь достаточно большой набор данных, иначе теряется сам смысл применения сложных инструментов. Для работы мы используем набор данных [League of Legends Diamond Ranked Games (10 min)](https://www.kaggle.com/datasets/bobbyscience/league-of-legends-diamond-ranked-games-10-min), который содержит  статистику 9879 рейтинговых игр (SOLO QUEUE) с высоким рейтингом ELO  (DIAMOND I - MASTER)за первые 10 минут игры. "
      ],
      "metadata": {
        "id": "4cZELt2JKvGH"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 237
        },
        "id": "eEl0Ds9YbISX",
        "outputId": "3a602c39-2917-4675-8375-20aa5d9a0887"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "       gameId  blueWins  blueWardsPlaced  blueWardsDestroyed  blueFirstBlood  \\\n",
              "0  4519157822         0               28                   2               1   \n",
              "1  4523371949         0               12                   1               0   \n",
              "2  4521474530         0               15                   0               0   \n",
              "\n",
              "   blueKills  blueDeaths  blueAssists  blueEliteMonsters  blueDragons  ...  \\\n",
              "0          9           6           11                  0            0  ...   \n",
              "1          5           5            5                  0            0  ...   \n",
              "2          7          11            4                  1            1  ...   \n",
              "\n",
              "   redTowersDestroyed  redTotalGold  redAvgLevel  redTotalExperience  \\\n",
              "0                   0         16567          6.8               17047   \n",
              "1                   1         17620          6.8               17438   \n",
              "2                   0         17285          6.8               17254   \n",
              "\n",
              "   redTotalMinionsKilled  redTotalJungleMinionsKilled  redGoldDiff  \\\n",
              "0                    197                           55         -643   \n",
              "1                    240                           52         2908   \n",
              "2                    203                           28         1172   \n",
              "\n",
              "   redExperienceDiff  redCSPerMin  redGoldPerMin  \n",
              "0                  8         19.7         1656.7  \n",
              "1               1173         24.0         1762.0  \n",
              "2               1033         20.3         1728.5  \n",
              "\n",
              "[3 rows x 40 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-53b11f7e-3931-4732-9662-40fbcfbf6c8b\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>gameId</th>\n",
              "      <th>blueWins</th>\n",
              "      <th>blueWardsPlaced</th>\n",
              "      <th>blueWardsDestroyed</th>\n",
              "      <th>blueFirstBlood</th>\n",
              "      <th>blueKills</th>\n",
              "      <th>blueDeaths</th>\n",
              "      <th>blueAssists</th>\n",
              "      <th>blueEliteMonsters</th>\n",
              "      <th>blueDragons</th>\n",
              "      <th>...</th>\n",
              "      <th>redTowersDestroyed</th>\n",
              "      <th>redTotalGold</th>\n",
              "      <th>redAvgLevel</th>\n",
              "      <th>redTotalExperience</th>\n",
              "      <th>redTotalMinionsKilled</th>\n",
              "      <th>redTotalJungleMinionsKilled</th>\n",
              "      <th>redGoldDiff</th>\n",
              "      <th>redExperienceDiff</th>\n",
              "      <th>redCSPerMin</th>\n",
              "      <th>redGoldPerMin</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>4519157822</td>\n",
              "      <td>0</td>\n",
              "      <td>28</td>\n",
              "      <td>2</td>\n",
              "      <td>1</td>\n",
              "      <td>9</td>\n",
              "      <td>6</td>\n",
              "      <td>11</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>16567</td>\n",
              "      <td>6.8</td>\n",
              "      <td>17047</td>\n",
              "      <td>197</td>\n",
              "      <td>55</td>\n",
              "      <td>-643</td>\n",
              "      <td>8</td>\n",
              "      <td>19.7</td>\n",
              "      <td>1656.7</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>4523371949</td>\n",
              "      <td>0</td>\n",
              "      <td>12</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>5</td>\n",
              "      <td>5</td>\n",
              "      <td>5</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>1</td>\n",
              "      <td>17620</td>\n",
              "      <td>6.8</td>\n",
              "      <td>17438</td>\n",
              "      <td>240</td>\n",
              "      <td>52</td>\n",
              "      <td>2908</td>\n",
              "      <td>1173</td>\n",
              "      <td>24.0</td>\n",
              "      <td>1762.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>4521474530</td>\n",
              "      <td>0</td>\n",
              "      <td>15</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>7</td>\n",
              "      <td>11</td>\n",
              "      <td>4</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>17285</td>\n",
              "      <td>6.8</td>\n",
              "      <td>17254</td>\n",
              "      <td>203</td>\n",
              "      <td>28</td>\n",
              "      <td>1172</td>\n",
              "      <td>1033</td>\n",
              "      <td>20.3</td>\n",
              "      <td>1728.5</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>3 rows × 40 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-53b11f7e-3931-4732-9662-40fbcfbf6c8b')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-53b11f7e-3931-4732-9662-40fbcfbf6c8b button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-53b11f7e-3931-4732-9662-40fbcfbf6c8b');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 2
        }
      ],
      "source": [
        "lol = pd.read_csv('https://raw.githubusercontent.com/vn322/Sport-University/main/esport/high_diamond_ranked_10min.csv')\n",
        "lol.head(3)"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "lol.columns"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NyWqpto8eXRQ",
        "outputId": "d69fe9f0-0cba-4653-87c0-c8ac9a593e35"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Index(['gameId', 'blueWins', 'blueWardsPlaced', 'blueWardsDestroyed',\n",
              "       'blueFirstBlood', 'blueKills', 'blueDeaths', 'blueAssists',\n",
              "       'blueEliteMonsters', 'blueDragons', 'blueHeralds',\n",
              "       'blueTowersDestroyed', 'blueTotalGold', 'blueAvgLevel',\n",
              "       'blueTotalExperience', 'blueTotalMinionsKilled',\n",
              "       'blueTotalJungleMinionsKilled', 'blueGoldDiff', 'blueExperienceDiff',\n",
              "       'blueCSPerMin', 'blueGoldPerMin', 'redWardsPlaced', 'redWardsDestroyed',\n",
              "       'redFirstBlood', 'redKills', 'redDeaths', 'redAssists',\n",
              "       'redEliteMonsters', 'redDragons', 'redHeralds', 'redTowersDestroyed',\n",
              "       'redTotalGold', 'redAvgLevel', 'redTotalExperience',\n",
              "       'redTotalMinionsKilled', 'redTotalJungleMinionsKilled', 'redGoldDiff',\n",
              "       'redExperienceDiff', 'redCSPerMin', 'redGoldPerMin'],\n",
              "      dtype='object')"
            ]
          },
          "metadata": {},
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Разделение данных"
      ],
      "metadata": {
        "id": "XB2aOs1OYMgK"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Разделение набора данных в Scikit-learn решено функцией `sklearn.model_selection.train_test_split`. Сначала нужно определить целевую (зависимую) переменную, у нас это будет индикатор `blueWins` принимающий значение \"1\" в случае победы команды синих, и \"0\" - в случае поражения.\n",
        "\n",
        "После определения целевой переменной необходимо разделить данные. Ведь нельзя оценить прогностическую эффективность модели на тех же данных, которые  использовались для её обучения. "
      ],
      "metadata": {
        "id": "WABi5wrYWum9"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "X=lol[['blueKills','blueDeaths','blueWardsPlaced', 'blueWardsDestroyed',\n",
        "       'blueTowersDestroyed', 'blueTotalGold', 'blueAssists','blueTotalMinionsKilled']]\n",
        "y=lol['blueWins']"
      ],
      "metadata": {
        "id": "F3Xhny59e19Y"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Чтобы объективно проверить точность  модели на данных, которые модель \"не видела\",  набор данных делят на две неравные части — тренировочный и тестовый набор . Тренировочный набор применяется для обучения модели, поэтому его делают большим, чтобы модель имела достаточно возможностей для обучения, а тестовый меньшим, достаточным для проверки. \n",
        "\n",
        "В следущем примере данные будут разделены в соотношении 70/30, т. е. 70% данных будут использоваться как\n",
        " данные для тренировки (обучения) модели, а 30% будут использоваться в качестве тестовых данных."
      ],
      "metadata": {
        "id": "4njsV6-lfhmz"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3,\n",
        "random_state=42)\n",
        "print(X_train.shape)\n",
        "print(X_test.shape)\n",
        "print(y_train.shape)\n",
        "print(y_test.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2u_DSVLNeNY4",
        "outputId": "e593aab6-dada-4938-b021-f2ee8fd40950"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(6915, 8)\n",
            "(2964, 8)\n",
            "(6915,)\n",
            "(2964,)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Проверим корректность  разделения данных по количеству и составу."
      ],
      "metadata": {
        "id": "3S2jgOtSV70z"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "X.iloc[[2]]"
      ],
      "metadata": {
        "id": "pH4jKJ35gFhQ",
        "outputId": "36c12c1d-cca9-4c4a-e9ba-c7ef043ad129",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 81
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "   blueKills  blueDeaths  blueWardsPlaced  blueWardsDestroyed  \\\n",
              "2          7          11               15                   0   \n",
              "\n",
              "   blueTowersDestroyed  blueTotalGold  blueAssists  blueTotalMinionsKilled  \n",
              "2                    0          16113            4                     186  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-a13cb579-8d0b-468a-a8e5-7f808e5e06e1\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>blueKills</th>\n",
              "      <th>blueDeaths</th>\n",
              "      <th>blueWardsPlaced</th>\n",
              "      <th>blueWardsDestroyed</th>\n",
              "      <th>blueTowersDestroyed</th>\n",
              "      <th>blueTotalGold</th>\n",
              "      <th>blueAssists</th>\n",
              "      <th>blueTotalMinionsKilled</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>7</td>\n",
              "      <td>11</td>\n",
              "      <td>15</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>16113</td>\n",
              "      <td>4</td>\n",
              "      <td>186</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-a13cb579-8d0b-468a-a8e5-7f808e5e06e1')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-a13cb579-8d0b-468a-a8e5-7f808e5e06e1 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-a13cb579-8d0b-468a-a8e5-7f808e5e06e1');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "y.iloc[[2]]"
      ],
      "metadata": {
        "id": "L9mzbbXmgyuv",
        "outputId": "3f1e3fc3-3729-4047-ccf4-7f09a380247e",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "2    0\n",
              "Name: blueWins, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Алгоритмы машинного обучения Scikit-Learn"
      ],
      "metadata": {
        "id": "GXNH7WI-eRDZ"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Более подробно применение каждого алгоритма мы разберём в следующем модуле, а сейчас сосредоточимся на рассмотрении возможностей именно библиотеки Scikit-Learn."
      ],
      "metadata": {
        "id": "Pw7gsxRQeWDk"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "У Scikit-learn единый способ использования всех методов.\n",
        "Для моделей (estimator object) доступны следующие методы.\n",
        "\n",
        "- `model.fit()` – настройка на данные (обучение)\n",
        "\n",
        "- `model.fit(X, y)` – для обучения с учителем (supervised learning)\n",
        "\n",
        "- `model.fit(X)` – для обучение без учителя (unsupervised learning)"
      ],
      "metadata": {
        "id": "G65Ie4qRikXY"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Для обучения с учителем:\n",
        "\n",
        "`model.predict(X_test)` – предсказать значения целевой переменной\n",
        "\n",
        "`model.predict_proba()` – выдать «степень уверенности» в ответе\n",
        "(вероятность) – для некоторых моделей\n",
        "\n",
        "`model.decision_function()` – решающая функция – для некоторых\n",
        "моделей\n",
        "\n",
        "`model.score()` – в большинстве моделей встроены методы оценки их\n",
        "качества работы\n",
        "\n",
        "`model.transform()` – для отбора признаков (feature selection)\n",
        "«сжимает» обучающую матрицу. Для регрессионных моделей и\n",
        "классификаторов (linear, RF и т.п.) выделяет наиболее\n",
        "информативные признаки"
      ],
      "metadata": {
        "id": "jgFilmEbjI0l"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Логистическая регрессия\n",
        "\n",
        "Начнём с классификации при помощи логистической регрессии. Обучим модель и предложим ей сделать прогноз на строке с индексом 2. В нашем наборе данных для данной строки целевая переменная равна 0, что значит поражение и прогноз так же равен 0. Однако, если мы подставим строку с индексом 1, то увидим, что модель даст ошибочный прогноз, поэтому необходимо понять с какой точностью делается прогноз и насколько он может быть надёжным."
      ],
      "metadata": {
        "id": "tJS07JR7X35k"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.linear_model import LogisticRegression\n",
        "classifier = LogisticRegression()\n",
        "classifier.fit(X_train, y_train)\n",
        "classifier.predict(X.iloc[[2]])"
      ],
      "metadata": {
        "id": "huO-Ydbrjq_5",
        "outputId": "3dc75b89-68a2-4201-b70d-2e84eab81e2c",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.8/dist-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
            "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
            "\n",
            "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
            "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
            "Please also refer to the documentation for alternative solver options:\n",
            "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
            "  n_iter_i = _check_optimize_result(\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([0])"
            ]
          },
          "metadata": {},
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Для оценки будем использовать штатный метод `score(X, y[, sample_weight])` который возвращает Accuracy - отношение точных прогнозов к общему их количеству или точнее\n",
        "\n",
        "$$\\text{Accuracy} = \\frac{TP+TN}{TP+TN+FP+FN}$$\n",
        "\n",
        "где \n",
        "\n",
        "TP - истинные положительные результаты (True Positive) прогноз совпал с реальностью\n",
        "\n",
        "TN - истинные отрицательные результаты, (True Negative) , прогноз совпал с реальностью\n",
        "\n",
        "FP - ложные положительные результаты (False Positive) ошибка 1-го рода, модель прогнозировала положительный результат, а на самом деле он отрицательный\n",
        "\n",
        "FN - ложные отрицательные результаты (False Negative) ошибка 2-го рода, модель прогнозировала отрицательный результат, но на самом деле он положительный"
      ],
      "metadata": {
        "id": "zO3LcEctqpHC"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "И в данных условиях для данной модели получим точность 0.7, что безусловно неплохой результат, но место для ошибки остаётся и желательно точнее определить сколько и каких ошибок ожидается при пользовании моделью."
      ],
      "metadata": {
        "id": "yTmSGys_ZLGc"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "classifier.score(X_test, y_test)"
      ],
      "metadata": {
        "id": "sULNGGiPj4o-",
        "outputId": "bff90c5b-03f0-4891-c38d-e8b182379253",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.7058029689608637"
            ]
          },
          "metadata": {},
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "С целью оценки прогноза используется матрица ошибок (confusion matrix) – таблица с 4 различными комбинациями прогнозируемых и фактических значений. Прогнозируемые значения описываются как положительные и отрицательные, а фактические – как истинные и ложные. Вообще матрица ошибок используется для оценки точности моделей в задачах классификации. Но многие другие задачи можно рассматривать как частный случай этой проблемы, поэтому confusion matrix актуальна и для измерения точности прогноза. Важно, что матрица ошибок позволяет оценить эффективность прогноза не только в качественном, но и в количественном выражении"
      ],
      "metadata": {
        "id": "ng62r24k1bv0"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import confusion_matrix\n",
        "y_pred = classifier.predict(X_test)\n",
        "confusion_matrix = confusion_matrix(y_test, y_pred)\n",
        "print(confusion_matrix)"
      ],
      "metadata": {
        "id": "JXiXZRwqnCY_",
        "outputId": "1eb295cc-41dc-474b-9ed3-7cd3d1e9b11d",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[1035  445]\n",
            " [ 427 1057]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Не всегда числа хорошо воспринимаются нашим сознанием  и поэтому, при возможности, желательно визуализировать результаты исследования."
      ],
      "metadata": {
        "id": "zgkpVn-vZ6jo"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import plot_confusion_matrix\n",
        "plot_confusion_matrix(classifier, X_test, y_test, cmap=plt.cm.Blues)\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "IfmesO2lzXjD",
        "outputId": "d9f07e78-5d76-4469-8227-ccd2ffdf62a4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 334
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.8/dist-packages/sklearn/utils/deprecation.py:87: FutureWarning: Function plot_confusion_matrix is deprecated; Function `plot_confusion_matrix` is deprecated in 1.0 and will be removed in 1.2. Use one of the class methods: ConfusionMatrixDisplay.from_predictions or ConfusionMatrixDisplay.from_estimator.\n",
            "  warnings.warn(msg, category=FutureWarning)\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "ROC — Receiver Operating Characteristic (рабочая характеристика приёмника). Впервые она была создана для использования радиолокационного обнаружения сигналов во время Второй мировой войны. США использовали ROC для повышения точности обнаружения японских самолетов с помощью радара. \n",
        "\n",
        "Матрица ошибок содержит только абсолютные числа. Однако, используя их, мы можем получить множество других метрик, основанных на процентных соотношениях. \n",
        "\n",
        "Так вводят: True Positive Rate (TPR) и False Positive Rate (FPR)\n",
        "\n",
        "True Positive Rate (TPR) показывает, какой процент среди всех positive верно предсказан моделью.\n",
        "$$TPR = TP / (TP + FN)$$\n",
        "\n",
        "False Positive Rate (FPR): какой процент среди всех negative неверно предсказан моделью.\n",
        "$$FPR = FP / (FP + TN)$$\n",
        "\n",
        "AUC или area under curve — это просто площадь под кривой ROC. А кривая ROC — это просто отношение TPR к FPR."
      ],
      "metadata": {
        "id": "-s2ZQMC7gYjd"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import roc_auc_score\n",
        "from sklearn.metrics import roc_curve\n",
        "logit_roc_auc = roc_auc_score(y_test, classifier.predict(X_test))\n",
        "fpr, tpr, thresholds = roc_curve(y_test, classifier.predict_proba(X_test)[:,1])\n",
        "plt.figure()\n",
        "plt.plot(fpr, tpr, label='Logistic Regression (area = %0.2f)' % logit_roc_auc)\n",
        "plt.plot([0, 1], [0, 1],'r--')\n",
        "plt.xlim([0.0, 1.0])\n",
        "plt.ylim([0.0, 1.05])\n",
        "plt.xlabel('False Positive Rate')\n",
        "plt.ylabel('True Positive Rate')\n",
        "plt.title('Receiver operating characteristic')\n",
        "plt.legend(loc=\"lower right\")\n",
        "plt.savefig('Log_ROC')\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "F7ttkf34oH0o",
        "outputId": "dcefc343-64ae-43ae-fa21-3ee564499b8c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "При использовании моделей классификации в машинном обучении есть несколько общих показателя, которые применяются для оценки качества модели:\n",
        "\n",
        "1. **precision** - процент правильных положительных прогнозов по отношению к общему количеству положительных прогнозов.\n",
        "\n",
        "2. **recall** - процент правильных положительных прогнозов по отношению к общему количеству фактических положительных результатов.\n",
        "\n",
        "\n",
        "$${\\displaystyle {\\begin{aligned}{\\text{Precision}}&={\\frac {TP}{TP+FN}}\\\\{\\text{Recall}}&={\\frac {TP}{TP+FN}}\\,\\end{aligned}}}$$\n",
        "3. **f1-score** - средневзвешенное гармоническое значение точности и полноты. Чем ближе к 1, тем лучше модель.\n",
        "\n",
        "$$\n",
        "{\\displaystyle F_1=2\\cdot {\\frac {\\mathrm {Precision} \\cdot \\mathrm {Recall} }{\\mathrm {Precision} +\\mathrm {Recall} }}}\n",
        "$$\n",
        "Используя эти три показателя, мы можем понять, насколько хорошо данная модель классификации способна предсказывать результаты для некоторой переменной отклика ."
      ],
      "metadata": {
        "id": "Pzg4LyGHix7X"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import classification_report\n",
        "print(classification_report(y_test, y_pred))"
      ],
      "metadata": {
        "id": "0WQv6CEqn7d3",
        "outputId": "74fe36e9-10fe-44ed-926c-846c647ebed1",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.71      0.70      0.70      1480\n",
            "           1       0.70      0.71      0.71      1484\n",
            "\n",
            "    accuracy                           0.71      2964\n",
            "   macro avg       0.71      0.71      0.71      2964\n",
            "weighted avg       0.71      0.71      0.71      2964\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Ансамблевые методы"
      ],
      "metadata": {
        "id": "xHwLbJCA2e1S"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Цель ансамблевых методов — объединить прогнозы нескольких базовых оценок, построенных с заданным алгоритмом обучения, чтобы улучшить надежность по сравнению с одной оценкой.\n",
        "\n",
        "Обычно выделяют два семейства ансамблевых методов:\n",
        "\n",
        "В методах усреднения главный принцип состоит в том, чтобы  усреднить прогнозы . В среднем, комбинированная оценка обычно лучше, чем любая из оценок с одной моделью, потому что ее дисперсия имеет тенденцию к уменьшению. \n",
        "\n",
        "Напротив, в методах с использованием бустинга оценки строятся последовательно, и каждая следующая пытается уменьшить смещение комбинированной оценки. \n"
      ],
      "metadata": {
        "id": "nHuPDJtUiHVH"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Используем популярный алгоритм классификации \"Решающее дерево\" и получим для него оценку штатным методом `score` отражающий среднюю точность (mean accuracy) равную 0.6740890688259109"
      ],
      "metadata": {
        "id": "_Ou5UORv5Lbw"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Решающее дерево"
      ],
      "metadata": {
        "id": "azrUblSP3UOV"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.tree import DecisionTreeClassifier\n",
        "tree = DecisionTreeClassifier(max_depth=10)\n",
        "tree.fit(X_train, y_train)\n",
        "print (tree)\n",
        "print (tree.score(X_test, y_test))\n",
        "print (tree.predict(X.iloc[[1]]))"
      ],
      "metadata": {
        "id": "dcL8WMbPkgWO",
        "outputId": "64c8de91-a2ab-4a5c-a168-5130e4850b40",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "DecisionTreeClassifier(max_depth=10)\n",
            "0.6724021592442645\n",
            "[0]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Случайный  лес"
      ],
      "metadata": {
        "id": "AgOKbPle3X50"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Используя же метод \"Случайного леса\", который прпедставляет из себя усреднённую оценку нескольких \"Решающих деревьев\" получим уже среднюю точность равную 0.7068151147098516"
      ],
      "metadata": {
        "id": "NIE56Nl66c9u"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.ensemble import RandomForestClassifier\n",
        "rf = RandomForestClassifier(random_state=42, n_estimators=100)\n",
        "rf.fit(X_train, y_train)\n"
      ],
      "metadata": {
        "id": "Z5NnKLFAiD59",
        "outputId": "59e1b225-0b01-44f6-fe81-730254a62ba4",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "RandomForestClassifier(random_state=42)"
            ]
          },
          "metadata": {},
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print (rf.score(X_test, y_test))\n",
        "print (rf.predict(X.iloc[[1]]))"
      ],
      "metadata": {
        "id": "rIyjkPt9g-w9",
        "outputId": "b24fb2d8-a396-499f-9b42-50b7effb58b6",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0.7068151147098516\n",
            "[0]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import classification_report\n",
        "y_pred = rf.predict(X_test)\n",
        "print(classification_report(y_test, y_pred))\n"
      ],
      "metadata": {
        "id": "p19wsFa5aacd",
        "outputId": "a4c92def-552f-4ea5-e837-cc1b181e4d00",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.70      0.72      0.71      1480\n",
            "           1       0.71      0.69      0.70      1484\n",
            "\n",
            "    accuracy                           0.71      2964\n",
            "   macro avg       0.71      0.71      0.71      2964\n",
            "weighted avg       0.71      0.71      0.71      2964\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Gradient boosting"
      ],
      "metadata": {
        "id": "faZHDM_B0nOA"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Gradient boosting — это метод машинного обучения, используемый в задачах как регрессии, так и классификации. Он создает модель прогнозирования в виде ансамбля более слабых моделей, которые обычно представляют собой деревья решений.  Обычно он превосходит случайный лес. Модель деревьев с градиентным усилением строится поэтапно, как и в других методах повышения, но она обобщает другие методы, позволяя оптимизировать произвольную дифференцируемую функцию потерь."
      ],
      "metadata": {
        "id": "K59XuT2k0N28"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.ensemble import GradientBoostingClassifier\n",
        "grb = GradientBoostingClassifier(n_estimators=100, learning_rate=1.0, max_depth=1, random_state=42).fit(X_train, y_train)\n",
        "grb.score(X_train, y_train)"
      ],
      "metadata": {
        "id": "Z1vWwiMsZ_b1",
        "outputId": "20d1e1b4-b834-40e0-bbca-38dbdd4b18de",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.7297180043383948"
            ]
          },
          "metadata": {},
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "С этим методом мы можем ещё немного улучшить среднюю точность и теперь она равна 0.7128879892037787"
      ],
      "metadata": {
        "id": "zcUGwzc67lTo"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "grb.score(X_test, y_test)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CKFtaH1xaIlb",
        "outputId": "71454f5c-7095-4e97-c690-a067a0045aa5"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.7128879892037787"
            ]
          },
          "metadata": {},
          "execution_count": 19
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "grb.predict(X.iloc[[1]])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "w0n9v5vnfyiU",
        "outputId": "93d95e0b-826d-4e53-8556-d2b7d7c15ea7"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([0])"
            ]
          },
          "metadata": {},
          "execution_count": 20
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import classification_report\n",
        "y_pred = grb.predict(X_test)\n",
        "print(classification_report(y_test, y_pred))"
      ],
      "metadata": {
        "id": "qe9lMa2k14gs",
        "outputId": "cff53578-1602-4c00-d76a-fc40b3256163",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.71      0.71      0.71      1480\n",
            "           1       0.71      0.71      0.71      1484\n",
            "\n",
            "    accuracy                           0.71      2964\n",
            "   macro avg       0.71      0.71      0.71      2964\n",
            "weighted avg       0.71      0.71      0.71      2964\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.neighbors import KNeighborsClassifier\n",
        "knn = KNeighborsClassifier(n_neighbors=5)\n",
        "knn.fit(X_train, y_train)\n",
        "print (knn)\n",
        "print (knn.score(X_test, y_test))"
      ],
      "metadata": {
        "id": "7ZwK87p_lewf",
        "outputId": "5a10ca9f-d132-4317-d123-f0d68f12735b",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "KNeighborsClassifier()\n",
            "0.6497975708502024\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Метод опорных векторов"
      ],
      "metadata": {
        "id": "Zs7A1yb25Cml"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Метод опорных векторов (support vector machine) — набор схожих алгоритмов обучения с учителем, использующихся для задач классификации и регрессионного анализа.  Особым свойством метода опорных векторов является непрерывное уменьшение эмпирической ошибки классификации и увеличение зазора, поэтому метод также известен как метод классификатора с максимальным зазором."
      ],
      "metadata": {
        "id": "aXTdg0CH4FWC"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn import svm\n",
        "clf = svm.SVC(kernel='linear', C=1).fit(X_train, y_train)\n",
        "clf.score(X_test, y_test)"
      ],
      "metadata": {
        "id": "fKsiUIm6alTF",
        "outputId": "fe974fae-e364-4fe3-9446-c19bb73d59ea",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.7186234817813765"
            ]
          },
          "metadata": {},
          "execution_count": 23
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import classification_report\n",
        "y_pred = clf.predict(X_test)\n",
        "print(classification_report(y_test, y_pred))"
      ],
      "metadata": {
        "outputId": "564617a2-9514-4c6b-f02d-21dd084b00ce",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "so1ioOqJbDaz"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.74      0.67      0.70      1480\n",
            "           1       0.70      0.77      0.73      1484\n",
            "\n",
            "    accuracy                           0.72      2964\n",
            "   macro avg       0.72      0.72      0.72      2964\n",
            "weighted avg       0.72      0.72      0.72      2964\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        " Стандартизация наборов данных является общим требованием для многих средств  машинного обучения, реализованных в scikit-learn; они могут не вполне корректно работать, если данные не распределены нормально.\n",
        "\n",
        "`StandardScaler` используется для изменения размера распределения значений так, чтобы среднее значение наблюдаемых значений было равно 0, а стандартное отклонение – 1. \n",
        "Этот метод часто является этапом препроцессинга  (предварительной обработки данных).\n",
        "\n",
        "Метод используется если характеристики  набора данных сильно различаются по размаху или, когда  измеряются в разных единицах измерения.\n"
      ],
      "metadata": {
        "id": "YppB5agE831I"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn import preprocessing\n",
        "\n",
        "scaler = preprocessing.StandardScaler().fit(X_train)\n",
        "X_train_transformed = scaler.transform(X_train)\n",
        "clf = svm.SVC(C=1).fit(X_train_transformed, y_train)\n",
        "X_test_transformed = scaler.transform(X_test)\n",
        "clf.score(X_test_transformed, y_test)"
      ],
      "metadata": {
        "id": "FGpxi_BHak6L",
        "outputId": "b92309c5-b025-4d73-e291-e56bfe0182ec",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.7139001349527665"
            ]
          },
          "metadata": {},
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "данные можно улучшать различными способами, нпраимер метод `get_dummies` позволяет из категориальных переменных получать так называемые \"фиктивные\", т.е представлять их числами для дальнейшей обработки моделью. \n",
        "\n",
        "Создадим категориальную переменную на основе нашего набора данных."
      ],
      "metadata": {
        "id": "VcKjhUIh5D6a"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "df = lol[['blueAvgLevel', 'blueAssists']]\n",
        "df['level'] = \"low\"\n",
        "condition = (lol['blueAvgLevel'] >= 7)\n",
        "df.loc[condition, 'level'] = \"high\"\n",
        "df.head()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 414
        },
        "id": "85TeTWSx0sZ4",
        "outputId": "f06b99c6-fcc6-409a-9731-761527cbaeeb"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "<ipython-input-56-e8cb7c5319f3>:2: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  df['level'] = \"low\"\n",
            "/usr/local/lib/python3.8/dist-packages/pandas/core/indexing.py:1817: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self._setitem_single_column(loc, value, pi)\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "   blueAvgLevel  blueAssists level\n",
              "0           6.6           11   low\n",
              "1           6.6            5   low\n",
              "2           6.4            4   low\n",
              "3           7.0            5  high\n",
              "4           7.0            6  high"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-3325c3eb-ee6c-4942-9237-5cb30e8003ed\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>blueAvgLevel</th>\n",
              "      <th>blueAssists</th>\n",
              "      <th>level</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>6.6</td>\n",
              "      <td>11</td>\n",
              "      <td>low</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>6.6</td>\n",
              "      <td>5</td>\n",
              "      <td>low</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>6.4</td>\n",
              "      <td>4</td>\n",
              "      <td>low</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>7.0</td>\n",
              "      <td>5</td>\n",
              "      <td>high</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>7.0</td>\n",
              "      <td>6</td>\n",
              "      <td>high</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-3325c3eb-ee6c-4942-9237-5cb30e8003ed')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-3325c3eb-ee6c-4942-9237-5cb30e8003ed button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-3325c3eb-ee6c-4942-9237-5cb30e8003ed');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 56
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Теперь превратим категориальные переменные в фиктивные и обратим внимание что во первых мы получили разреженную матрицу, где большое количество нулей, зато столбцы однозначно отражают наличие или отсутствие нужного признака."
      ],
      "metadata": {
        "id": "6eICjdzM6agq"
      }
    },
    {
      "cell_type": "code",
      "source": [
        " pd.get_dummies(df).head()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "-jPBwvrm2eUG",
        "outputId": "02d312d6-c1d6-47b4-c0d9-3f5d96697dd4"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "   blueAvgLevel  blueAssists  level_high  level_low\n",
              "0           6.6           11           0          1\n",
              "1           6.6            5           0          1\n",
              "2           6.4            4           0          1\n",
              "3           7.0            5           1          0\n",
              "4           7.0            6           1          0"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-78ebf868-d9cd-4f09-b763-f432219002fa\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>blueAvgLevel</th>\n",
              "      <th>blueAssists</th>\n",
              "      <th>level_high</th>\n",
              "      <th>level_low</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>6.6</td>\n",
              "      <td>11</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>6.6</td>\n",
              "      <td>5</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>6.4</td>\n",
              "      <td>4</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>7.0</td>\n",
              "      <td>5</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>7.0</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-78ebf868-d9cd-4f09-b763-f432219002fa')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-78ebf868-d9cd-4f09-b763-f432219002fa button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-78ebf868-d9cd-4f09-b763-f432219002fa');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 57
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "В библиотеке Scikit-learn существуют и другие способы подготовки данных,  снижения их размерности и других преобразований."
      ],
      "metadata": {
        "id": "4B4vSQpOzteZ"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "`Pipeline` используется для объединения операций предварительной обработки (например, масштабирования данных или one-hot-encoding) с моделью машинного обучения типа классификаторов. Его использование позволяет избежать ошибок и сокращает временные издержки."
      ],
      "metadata": {
        "id": "uvGMg8et95JR"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.ensemble import RandomForestClassifier\n",
        "from sklearn.pipeline import Pipeline\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "\n",
        "# Chain together scaling the variables with the model\n",
        "pipe = Pipeline([('scaler', StandardScaler()), ('rf', RandomForestClassifier())])\n",
        "pipe.fit(X_train, y_train)\n",
        "\n",
        "pipe.score(X_test, y_test)"
      ],
      "metadata": {
        "id": "S9gn1h0sfqaH",
        "outputId": "0f8890da-a0a7-4b6b-d67a-1cb66c64ae11",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.7064777327935222"
            ]
          },
          "metadata": {},
          "execution_count": 55
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.linear_model import LogisticRegression\n",
        "from sklearn.pipeline import make_pipeline\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "\n",
        "pipe = make_pipeline(StandardScaler(), LogisticRegression())\n",
        "pipe.fit(X_train, y_train)  # apply scaling on training data\n",
        "\n",
        "pipe.score(X_test, y_test)"
      ],
      "metadata": {
        "id": "l6wHCTsQak0j",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "8bcf6111-4806-477d-84ec-9428e0e5f95a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.7139001349527665"
            ]
          },
          "metadata": {},
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Кластерный анализ\n",
        "\n",
        "Кроме моделей регрессии и классификации в библиотеке Scikit-learn присутствуют методы для кластерного анализа."
      ],
      "metadata": {
        "id": "RHo2zTHKLPqb"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Алгоритм KMeans группирует данные, пытаясь разделить выборки на n групп с одинаковой дисперсией, сводя к минимуму критерий, известный как инерция или сумма квадратов внутри кластера (см. ниже). Этот алгоритм требует указания количества кластеров. Он хорошо масштабируется для большого количества образцов и используется в самых разных областях применения в самых разных областях.\n",
        "\n",
        "В нашем наборе данных нет категориальных переменных, но мы включим этот шаг, чтобы в дальнейшнм вы могли использовать код для своих проектов."
      ],
      "metadata": {
        "id": "fSkIbEzjLUzF"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "df_kmeans = pd.get_dummies(lol, drop_first=True)\n",
        "df_kmeans.head()"
      ],
      "metadata": {
        "id": "JD_vtFUrPI53",
        "outputId": "81d23bca-780e-4c0b-ad85-37f13bd6a47b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 300
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "       gameId  blueWins  blueWardsPlaced  blueWardsDestroyed  blueFirstBlood  \\\n",
              "0  4519157822         0               28                   2               1   \n",
              "1  4523371949         0               12                   1               0   \n",
              "2  4521474530         0               15                   0               0   \n",
              "3  4524384067         0               43                   1               0   \n",
              "4  4436033771         0               75                   4               0   \n",
              "\n",
              "   blueKills  blueDeaths  blueAssists  blueEliteMonsters  blueDragons  ...  \\\n",
              "0          9           6           11                  0            0  ...   \n",
              "1          5           5            5                  0            0  ...   \n",
              "2          7          11            4                  1            1  ...   \n",
              "3          4           5            5                  1            0  ...   \n",
              "4          6           6            6                  0            0  ...   \n",
              "\n",
              "   redTotalGold  redAvgLevel  redTotalExperience  redTotalMinionsKilled  \\\n",
              "0         16567          6.8               17047                    197   \n",
              "1         17620          6.8               17438                    240   \n",
              "2         17285          6.8               17254                    203   \n",
              "3         16478          7.0               17961                    235   \n",
              "4         17404          7.0               18313                    225   \n",
              "\n",
              "   redTotalJungleMinionsKilled  redGoldDiff  redExperienceDiff  redCSPerMin  \\\n",
              "0                           55         -643                  8         19.7   \n",
              "1                           52         2908               1173         24.0   \n",
              "2                           28         1172               1033         20.3   \n",
              "3                           47         1321                  7         23.5   \n",
              "4                           67         1004               -230         22.5   \n",
              "\n",
              "   redGoldPerMin  Clusters  \n",
              "0         1656.7         2  \n",
              "1         1762.0         0  \n",
              "2         1728.5         1  \n",
              "3         1647.8         0  \n",
              "4         1740.4         3  \n",
              "\n",
              "[5 rows x 41 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-540c6061-568d-49f7-8d24-8b0d5d02b026\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>gameId</th>\n",
              "      <th>blueWins</th>\n",
              "      <th>blueWardsPlaced</th>\n",
              "      <th>blueWardsDestroyed</th>\n",
              "      <th>blueFirstBlood</th>\n",
              "      <th>blueKills</th>\n",
              "      <th>blueDeaths</th>\n",
              "      <th>blueAssists</th>\n",
              "      <th>blueEliteMonsters</th>\n",
              "      <th>blueDragons</th>\n",
              "      <th>...</th>\n",
              "      <th>redTotalGold</th>\n",
              "      <th>redAvgLevel</th>\n",
              "      <th>redTotalExperience</th>\n",
              "      <th>redTotalMinionsKilled</th>\n",
              "      <th>redTotalJungleMinionsKilled</th>\n",
              "      <th>redGoldDiff</th>\n",
              "      <th>redExperienceDiff</th>\n",
              "      <th>redCSPerMin</th>\n",
              "      <th>redGoldPerMin</th>\n",
              "      <th>Clusters</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>4519157822</td>\n",
              "      <td>0</td>\n",
              "      <td>28</td>\n",
              "      <td>2</td>\n",
              "      <td>1</td>\n",
              "      <td>9</td>\n",
              "      <td>6</td>\n",
              "      <td>11</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>16567</td>\n",
              "      <td>6.8</td>\n",
              "      <td>17047</td>\n",
              "      <td>197</td>\n",
              "      <td>55</td>\n",
              "      <td>-643</td>\n",
              "      <td>8</td>\n",
              "      <td>19.7</td>\n",
              "      <td>1656.7</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>4523371949</td>\n",
              "      <td>0</td>\n",
              "      <td>12</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>5</td>\n",
              "      <td>5</td>\n",
              "      <td>5</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>17620</td>\n",
              "      <td>6.8</td>\n",
              "      <td>17438</td>\n",
              "      <td>240</td>\n",
              "      <td>52</td>\n",
              "      <td>2908</td>\n",
              "      <td>1173</td>\n",
              "      <td>24.0</td>\n",
              "      <td>1762.0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>4521474530</td>\n",
              "      <td>0</td>\n",
              "      <td>15</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>7</td>\n",
              "      <td>11</td>\n",
              "      <td>4</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>...</td>\n",
              "      <td>17285</td>\n",
              "      <td>6.8</td>\n",
              "      <td>17254</td>\n",
              "      <td>203</td>\n",
              "      <td>28</td>\n",
              "      <td>1172</td>\n",
              "      <td>1033</td>\n",
              "      <td>20.3</td>\n",
              "      <td>1728.5</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>4524384067</td>\n",
              "      <td>0</td>\n",
              "      <td>43</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>4</td>\n",
              "      <td>5</td>\n",
              "      <td>5</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>16478</td>\n",
              "      <td>7.0</td>\n",
              "      <td>17961</td>\n",
              "      <td>235</td>\n",
              "      <td>47</td>\n",
              "      <td>1321</td>\n",
              "      <td>7</td>\n",
              "      <td>23.5</td>\n",
              "      <td>1647.8</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>4436033771</td>\n",
              "      <td>0</td>\n",
              "      <td>75</td>\n",
              "      <td>4</td>\n",
              "      <td>0</td>\n",
              "      <td>6</td>\n",
              "      <td>6</td>\n",
              "      <td>6</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>17404</td>\n",
              "      <td>7.0</td>\n",
              "      <td>18313</td>\n",
              "      <td>225</td>\n",
              "      <td>67</td>\n",
              "      <td>1004</td>\n",
              "      <td>-230</td>\n",
              "      <td>22.5</td>\n",
              "      <td>1740.4</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>5 rows × 41 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-540c6061-568d-49f7-8d24-8b0d5d02b026')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-540c6061-568d-49f7-8d24-8b0d5d02b026 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-540c6061-568d-49f7-8d24-8b0d5d02b026');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 65
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Стандартизируем наши данные и обучим модель"
      ],
      "metadata": {
        "id": "UgO8xInA9ZdN"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.preprocessing import StandardScaler\n",
        "scaler = StandardScaler()\n",
        "scaled_df_kmeans = scaler.fit_transform(df_kmeans)"
      ],
      "metadata": {
        "id": "IoVeMqzbPXWt"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.cluster import KMeans\n",
        "kmeans_model = KMeans(n_clusters=3)\n",
        "clusters = kmeans_model.fit_predict(df_kmeans)\n",
        "df_kmeans.insert(df_kmeans.columns.get_loc(\"blueWins\"), \"Cluster\", clusters)\n",
        "df_kmeans.head(3)"
      ],
      "metadata": {
        "id": "Qc-hWweEPb51",
        "outputId": "19387341-498d-459c-aad3-59c7dac49279",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 237
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "       gameId  Cluster  blueWins  blueWardsPlaced  blueWardsDestroyed  \\\n",
              "0  4519157822        0         0               28                   2   \n",
              "1  4523371949        0         0               12                   1   \n",
              "2  4521474530        0         0               15                   0   \n",
              "\n",
              "   blueFirstBlood  blueKills  blueDeaths  blueAssists  blueEliteMonsters  ...  \\\n",
              "0               1          9           6           11                  0  ...   \n",
              "1               0          5           5            5                  0  ...   \n",
              "2               0          7          11            4                  1  ...   \n",
              "\n",
              "   redTotalGold  redAvgLevel  redTotalExperience  redTotalMinionsKilled  \\\n",
              "0         16567          6.8               17047                    197   \n",
              "1         17620          6.8               17438                    240   \n",
              "2         17285          6.8               17254                    203   \n",
              "\n",
              "   redTotalJungleMinionsKilled  redGoldDiff  redExperienceDiff  redCSPerMin  \\\n",
              "0                           55         -643                  8         19.7   \n",
              "1                           52         2908               1173         24.0   \n",
              "2                           28         1172               1033         20.3   \n",
              "\n",
              "   redGoldPerMin  Clusters  \n",
              "0         1656.7         2  \n",
              "1         1762.0         0  \n",
              "2         1728.5         1  \n",
              "\n",
              "[3 rows x 42 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-40225ef6-8c3b-4188-8375-b373e162f87f\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>gameId</th>\n",
              "      <th>Cluster</th>\n",
              "      <th>blueWins</th>\n",
              "      <th>blueWardsPlaced</th>\n",
              "      <th>blueWardsDestroyed</th>\n",
              "      <th>blueFirstBlood</th>\n",
              "      <th>blueKills</th>\n",
              "      <th>blueDeaths</th>\n",
              "      <th>blueAssists</th>\n",
              "      <th>blueEliteMonsters</th>\n",
              "      <th>...</th>\n",
              "      <th>redTotalGold</th>\n",
              "      <th>redAvgLevel</th>\n",
              "      <th>redTotalExperience</th>\n",
              "      <th>redTotalMinionsKilled</th>\n",
              "      <th>redTotalJungleMinionsKilled</th>\n",
              "      <th>redGoldDiff</th>\n",
              "      <th>redExperienceDiff</th>\n",
              "      <th>redCSPerMin</th>\n",
              "      <th>redGoldPerMin</th>\n",
              "      <th>Clusters</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>4519157822</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>28</td>\n",
              "      <td>2</td>\n",
              "      <td>1</td>\n",
              "      <td>9</td>\n",
              "      <td>6</td>\n",
              "      <td>11</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>16567</td>\n",
              "      <td>6.8</td>\n",
              "      <td>17047</td>\n",
              "      <td>197</td>\n",
              "      <td>55</td>\n",
              "      <td>-643</td>\n",
              "      <td>8</td>\n",
              "      <td>19.7</td>\n",
              "      <td>1656.7</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>4523371949</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>12</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>5</td>\n",
              "      <td>5</td>\n",
              "      <td>5</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>17620</td>\n",
              "      <td>6.8</td>\n",
              "      <td>17438</td>\n",
              "      <td>240</td>\n",
              "      <td>52</td>\n",
              "      <td>2908</td>\n",
              "      <td>1173</td>\n",
              "      <td>24.0</td>\n",
              "      <td>1762.0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>4521474530</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>15</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>7</td>\n",
              "      <td>11</td>\n",
              "      <td>4</td>\n",
              "      <td>1</td>\n",
              "      <td>...</td>\n",
              "      <td>17285</td>\n",
              "      <td>6.8</td>\n",
              "      <td>17254</td>\n",
              "      <td>203</td>\n",
              "      <td>28</td>\n",
              "      <td>1172</td>\n",
              "      <td>1033</td>\n",
              "      <td>20.3</td>\n",
              "      <td>1728.5</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>3 rows × 42 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-40225ef6-8c3b-4188-8375-b373e162f87f')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-40225ef6-8c3b-4188-8375-b373e162f87f button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-40225ef6-8c3b-4188-8375-b373e162f87f');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 67
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Метод локтя (Elbow Rule) — это распространенный способ найти оптимальное количество кластеров в наборе данных.\n",
        "Ось x – количество кластеров k, y – сумма квадрат расстояний между точками в кластере (SSD) (насколько точки данных удалены от центра тяжести кластера)\n"
      ],
      "metadata": {
        "id": "PaAuVDve-kBV"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "ssd = []\n",
        "for k in range(2, 9):\n",
        "    kmeans_model = KMeans(n_clusters=k)\n",
        "    kmeans_model.fit(df_kmeans)\n",
        "    ssd.append(kmeans_model.inertia_)\n",
        "plt.figure(figsize=(6, 4), dpi=100)\n",
        "plt.plot(range(2, 9), ssd, color=\"green\", marker=\"o\")\n",
        "plt.xlabel(\"Number of clusters (K)\")\n",
        "plt.ylabel(\"SSD for K\")\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "kX1kmrFUQFHv",
        "outputId": "3a219b61-35b3-4204-f52c-5f4468a3902c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 397
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 600x400 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "km.cluster_centers_"
      ],
      "metadata": {
        "id": "qakxitNRbUaC",
        "outputId": "ed7532c8-e8b0-4c47-a5b4-8bf1e2d1ff23",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[ 3.40586531,  4.34902245],\n",
              "       [ 5.07393266,  9.25060743],\n",
              "       [10.61799065,  7.10689252],\n",
              "       [ 7.48335975,  3.88431062]])"
            ]
          },
          "metadata": {},
          "execution_count": 69
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "lol['Clusters'] = km.labels_\n",
        "sns.scatterplot(x=\"blueKills\", y=\"blueDeaths\",hue = 'Clusters',  data=lol, palette='viridis')"
      ],
      "metadata": {
        "id": "snICEuuDbaz7",
        "outputId": "1c94fd08-f81b-477b-9b73-b39b8947451f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<AxesSubplot:xlabel='blueKills', ylabel='blueDeaths'>"
            ]
          },
          "metadata": {},
          "execution_count": 70
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Задание"
      ],
      "metadata": {
        "id": "a0ozQ6QW_qAE"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Третье исследование на собранном ранее наборе данных (он может быть дополнен или собран новый)\n",
        "\n",
        "- произвести стандартизацию данных\n",
        "- при необходимости преобразовать категориальные данные\n",
        "- разделить набор данных на тренировочный и тестовый\n",
        "- использовать на выбор любой метод машинного обучения из библиотеки Scikit-learn\n",
        "- описать каждый шаг исследования дав пояснение и обоснование\n"
      ],
      "metadata": {
        "id": "nXe0HzFWCm9O"
      }
    }
  ]
}