clustering_with_kmeans_and_dbscan.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": []
    },
    {
      "cell_type": "code",
      "execution_count": 111,
      "metadata": {},
      "outputs": [],
      "source": [
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "import pandas as pd"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## Exercício 1 - Dataset Digits do sklearn"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Acesso: https://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_digits.html#sklearn.datasets.load_digits"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 112,
      "metadata": {},
      "outputs": [],
      "source": [
        "# 1) Importar o pacote \"sklearn.datasets\" e o \"load_digits\"\n",
        "from sklearn.datasets import load_digits"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 113,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/plain": [
              "dict_keys(['data', 'target', 'frame', 'feature_names', 'target_names', 'images', 'DESCR'])"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# 2) Carregar o dataset através do método: load_digits()\n",
        "df_digits = load_digits()\n",
        "# 3) Observe as keys do dataset usando o método \"keys\"\n",
        "display(df_digits.keys())"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 114,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/plain": [
              "array([[ 0.,  0.,  5., ...,  0.,  0.,  0.],\n",
              "       [ 0.,  0.,  0., ..., 10.,  0.,  0.],\n",
              "       [ 0.,  0.,  0., ..., 16.,  9.,  0.],\n",
              "       ...,\n",
              "       [ 0.,  0.,  1., ...,  6.,  0.,  0.],\n",
              "       [ 0.,  0.,  2., ..., 12.,  0.,  0.],\n",
              "       [ 0.,  0., 10., ..., 12.,  1.,  0.]])"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "text/plain": [
              "array([0, 1, 2, ..., 8, 9, 8])"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# 4) A chave \"data\" são as features e a chave \"target\" é o y. Separe os dados em 2 variáveis diferentes\n",
        "X, y = df_digits[\"data\"], df_digits[\"target\"]\n",
        "display(X)\n",
        "display(y)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 115,
      "metadata": {},
      "outputs": [],
      "source": [
        "# 5) Separe o conjunto de dados em treinamento e teste usando o método: \"train_test_split\"\n",
        "from sklearn.model_selection import train_test_split\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 116,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/plain": [
              "{0, 1, 2, 3, 4, 5, 6, 7, 8, 9}"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "text/html": [
              "<style>#sk-container-id-14 {color: black;}#sk-container-id-14 pre{padding: 0;}#sk-container-id-14 div.sk-toggleable {background-color: white;}#sk-container-id-14 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-14 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-14 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-14 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-14 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-14 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-14 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-14 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-14 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-14 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-14 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-14 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-14 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-14 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-14 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-14 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-14 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-14 div.sk-item {position: relative;z-index: 1;}#sk-container-id-14 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-14 div.sk-item::before, #sk-container-id-14 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-14 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-14 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-14 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-14 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-14 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-14 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-14 div.sk-label-container {text-align: center;}#sk-container-id-14 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-14 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-14\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>KMeans(n_clusters=10, n_init=10, random_state=42)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-16\" type=\"checkbox\" checked><label for=\"sk-estimator-id-16\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">KMeans</label><div class=\"sk-toggleable__content\"><pre>KMeans(n_clusters=10, n_init=10, random_state=42)</pre></div></div></div></div></div>"
            ],
            "text/plain": [
              "KMeans(n_clusters=10, n_init=10, random_state=42)"
            ]
          },
          "execution_count": 116,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# 6) Treinar o K-Means, sendo o K seja igual ao número de classes do dataset\n",
        "from sklearn.cluster import KMeans\n",
        "\n",
        "myset = set(y_train)\n",
        "clusters = len(myset)\n",
        "display(myset)\n",
        "\n",
        "model = KMeans(n_clusters=clusters, n_init=10, random_state=42)\n",
        "model.fit(X_train)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 117,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/plain": [
              "{0: 4, 1: 1, 2: 3, 3: 7, 4: 1, 5: 6, 6: 9, 7: 5, 8: 2, 9: 0}"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "from collections import Counter\n",
        "\n",
        "# Pegar os labels dos padrões de Treinamento\n",
        "labels = model.labels_\n",
        "\n",
        "map_labels = []\n",
        "\n",
        "for i in range(clusters):\n",
        "  map_labels.append([])\n",
        "\n",
        "new_y_train = list(y_train)\n",
        "\n",
        "for i in range(len(y_train)):\n",
        "  for c in range(clusters):\n",
        "    if labels[i] == c:\n",
        "      map_labels[c].append(new_y_train[i])\n",
        "\n",
        "# Criar dicionário com os labells a serem mapeados\n",
        "mapping = {}\n",
        "\n",
        "for i in range(clusters):\n",
        "  final = Counter(map_labels[i]) # contar a classe que mais aparece\n",
        "  value = final.most_common(1)[0][0] # retorna a classe com maior frequência\n",
        "  mapping[i] = value\n",
        "\n",
        "display(mapping)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 118,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "80%\n",
            "[6, 9, 3, 7, 2, 2, 5, 3, 5, 2, 1, 1, 4, 0, 4, 2, 3, 7, 9, 9, 4, 3, 9, 7, 5, 6, 3, 5, 6, 3, 4, 9, 2, 4, 4, 6, 9, 4, 7, 6, 6, 9, 1, 3, 6, 1, 3, 0, 6, 5, 5, 1, 9, 5, 6, 0, 9, 0, 0, 1, 0, 4, 5, 2, 4, 9, 7, 0, 7, 9, 9, 9, 5, 7, 7, 0, 4, 5, 9, 9, 9, 0, 2, 3, 9, 0, 6, 4, 4, 9, 1, 2, 1, 3, 9, 2, 9, 4, 4, 7, 4, 3, 5, 3, 1, 1, 5, 1, 4, 2, 7, 7, 4, 4, 1, 9, 1, 7, 9, 7, 2, 6, 9, 4, 0, 7, 2, 7, 5, 1, 1, 5, 7, 7, 0, 6, 6, 4, 2, 1, 0, 9, 4, 6, 9, 1, 6, 9, 0, 9, 9, 6, 6, 0, 6, 4, 9, 9, 3, 1, 7, 2, 9, 0, 4, 5, 1, 6, 5, 7, 9, 1, 4, 2, 1, 3, 7, 7, 2, 2, 3, 9, 9, 0, 3, 2, 3, 5, 6, 9, 9, 4, 1, 5, 4, 2, 3, 6, 4, 1, 9, 9, 5, 7, 1, 9, 4, 1, 1, 5, 4, 4, 9, 6, 1, 9, 6, 0, 4, 5, 2, 7, 1, 6, 4, 5, 6, 0, 3, 2, 3, 6, 7, 1, 9, 2, 4, 7, 6, 5, 1, 9, 5, 1, 0, 2, 1, 1, 7, 5, 7, 6, 2, 2, 2, 3, 4, 9, 1, 3, 6, 0, 9, 7, 7, 0, 1, 0, 4, 9, 1, 5, 3, 6, 0, 4, 2, 0, 0, 3, 6, 9, 9, 7, 3, 5, 9, 9, 9, 1, 5, 3, 3, 1, 0, 5, 1, 3, 4, 0, 2, 4, 6, 4, 3, 4, 9, 0, 5, 2, 2, 3, 1, 4, 1, 1, 7, 0, 1, 5, 2, 1, 2, 9, 7, 0, 6, 4, 1, 1, 5, 1, 1, 4, 5, 1, 7, 9, 1, 6, 0, 6, 2, 0, 7, 9, 1, 9, 5, 2, 7, 7, 1, 9, 7, 4, 3, 7, 3, 9]\n",
            "[6, 9, 3, 7, 2, 1, 5, 2, 5, 2, 1, 9, 4, 0, 4, 2, 3, 7, 8, 8, 4, 3, 9, 7, 5, 6, 3, 5, 6, 3, 4, 9, 1, 4, 4, 6, 9, 4, 7, 6, 6, 9, 1, 3, 6, 1, 3, 0, 6, 5, 5, 1, 9, 5, 6, 0, 9, 0, 0, 1, 0, 4, 5, 2, 4, 5, 7, 0, 7, 5, 9, 5, 5, 4, 7, 0, 4, 5, 5, 9, 9, 0, 2, 3, 8, 0, 6, 4, 4, 9, 1, 2, 8, 3, 5, 2, 9, 0, 4, 4, 4, 3, 5, 3, 1, 3, 5, 9, 4, 2, 7, 7, 4, 4, 1, 9, 2, 7, 8, 7, 2, 6, 9, 4, 0, 7, 2, 7, 5, 8, 7, 5, 7, 7, 0, 6, 6, 4, 2, 8, 0, 9, 4, 6, 9, 9, 6, 9, 0, 3, 5, 6, 6, 0, 6, 4, 3, 9, 3, 9, 7, 2, 9, 0, 4, 5, 3, 6, 5, 9, 9, 8, 4, 2, 1, 3, 7, 7, 2, 2, 3, 9, 8, 0, 3, 2, 2, 5, 6, 9, 9, 4, 1, 5, 4, 2, 3, 6, 4, 8, 5, 9, 5, 7, 8, 9, 4, 8, 1, 5, 4, 4, 9, 6, 1, 8, 6, 0, 4, 5, 2, 7, 4, 6, 4, 5, 6, 0, 3, 2, 3, 6, 7, 1, 5, 1, 4, 7, 6, 8, 8, 5, 5, 1, 6, 2, 8, 8, 9, 9, 7, 6, 2, 2, 2, 3, 4, 8, 8, 3, 6, 0, 9, 7, 7, 0, 1, 0, 4, 5, 1, 5, 3, 6, 0, 4, 1, 0, 0, 3, 6, 5, 9, 7, 3, 5, 5, 9, 9, 8, 5, 3, 3, 2, 0, 5, 8, 3, 4, 0, 2, 4, 6, 4, 3, 4, 5, 0, 5, 2, 1, 3, 1, 4, 1, 1, 7, 0, 1, 5, 2, 1, 2, 8, 7, 0, 6, 4, 8, 8, 5, 1, 8, 4, 5, 8, 7, 9, 8, 5, 0, 6, 2, 0, 7, 9, 8, 9, 5, 2, 7, 7, 1, 8, 7, 4, 3, 8, 3, 5]\n"
          ]
        }
      ],
      "source": [
        "# 7) Mostrar a taxa de acerto do K-Means\n",
        "from sklearn import metrics\n",
        "\n",
        "result = model.predict(X_test)\n",
        "result = [mapping[i] for i in result]\n",
        "\n",
        "acc = metrics.accuracy_score(result, y_test)\n",
        "show = round(acc * 100)\n",
        "print(\"{}%\".format(show))\n",
        "\n",
        "print(list(result))\n",
        "print(list(y_test))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 119,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "2159057.2910406226\n",
            "1914619.6175501032\n",
            "1730182.260086908\n",
            "1612286.499390834\n",
            "1497595.4364257318\n",
            "1404969.3162221701\n",
            "1334825.6614009049\n",
            "1265263.8193929233\n",
            "1202278.336860986\n",
            "1165230.6872280112\n",
            "1131908.3396780784\n",
            "1099103.8677018434\n",
            "1078813.895604824\n",
            "1046289.720699995\n",
            "1019237.5443816143\n",
            "1007585.753065212\n",
            "989988.4644368881\n",
            "977897.2743826497\n",
            "959523.1355938826\n",
            "949867.9560077593\n",
            "933496.9512815909\n",
            "921084.7584645732\n",
            "904790.2719855744\n",
            "901654.095831646\n",
            "888493.0390309177\n",
            "873082.2164798807\n",
            "871385.324239637\n",
            "855300.4285150114\n",
            "845058.7576464504\n",
            "840126.224781521\n",
            "834088.3220536136\n",
            "822814.2397684435\n",
            "809064.3840358322\n",
            "811801.3203442268\n",
            "801097.9585294328\n",
            "797676.9957126256\n",
            "785761.8610594508\n",
            "777327.7042629528\n",
            "772811.5344347985\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# 8) Calcular a curva do cotovelo para verificar qual o número ideal de clusters para esse dataset\n",
        "sse = {}\n",
        "for k in range(1, 40):\n",
        "    kmeans = KMeans(n_clusters=k, n_init=10, max_iter=1000).fit(X)\n",
        "    sse[k] = kmeans.inertia_\n",
        "    print(sse[k])\n",
        "\n",
        "plt.figure()\n",
        "plt.plot(list(sse.keys()), list(sse.values()))\n",
        "plt.xlabel(\"Número de clusters\")\n",
        "plt.ylabel(\"SSE\")\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 120,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Valor do coeficiente de Silhouetter: 0.164\n"
          ]
        }
      ],
      "source": [
        "# 9) Calcular o Coeficiente de Silhouette para o K-Means treinado em (7)\n",
        "score = metrics.silhouette_score(X, km.labels_, metric='euclidean')\n",
        "print('Valor do coeficiente de Silhouetter: %.3f' % score)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 121,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "{'init': 'random', 'n_clusters': 14, 'n_init': 13}\n"
          ]
        }
      ],
      "source": [
        "# 10) É possível melhorar o Coeficiente de Silhouette? Altere o K-Means para conseguir um valor melhor do Coeficiente de Silhouette.\n",
        "from sklearn.model_selection import GridSearchCV\n",
        "\n",
        "param_grid = {'n_clusters': np.arange(8, 15), 'init': ['k-means++', 'random'], 'n_init': np.arange(5, 15)}\n",
        "grid = GridSearchCV(KMeans(), param_grid, cv=5)\n",
        "grid.fit(X)\n",
        "print(grid.best_params_)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 122,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Valor do coeficiente de Silhouetter: 0.183\n"
          ]
        }
      ],
      "source": [
        "best_model = grid.best_estimator_\n",
        "best_model.fit(X)\n",
        "score = metrics.silhouette_score(X, best_model.labels_, metric='euclidean')\n",
        "print('Valor do coeficiente de Silhouetter: %.3f' % score)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## Exercício 2 - Dataset Wine do sklearn\n",
        "Acesso: https://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_wine.html#sklearn.datasets.load_wine"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 123,
      "metadata": {},
      "outputs": [],
      "source": [
        "# 1) Importar o pacote \"sklearn.datasets\" e o \"load_wine\"\n",
        "from sklearn.datasets import load_wine"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 124,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/plain": [
              "dict_keys(['data', 'target', 'frame', 'target_names', 'DESCR', 'feature_names'])"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# 2) Carregar o dataset através do método: load_wine()\n",
        "df_wine = load_wine()\n",
        "# 3) Observe as keys do dataset usando o método \"keys\"\n",
        "display(df_wine.keys())"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 125,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/plain": [
              "array([[1.423e+01, 1.710e+00, 2.430e+00, ..., 1.040e+00, 3.920e+00,\n",
              "        1.065e+03],\n",
              "       [1.320e+01, 1.780e+00, 2.140e+00, ..., 1.050e+00, 3.400e+00,\n",
              "        1.050e+03],\n",
              "       [1.316e+01, 2.360e+00, 2.670e+00, ..., 1.030e+00, 3.170e+00,\n",
              "        1.185e+03],\n",
              "       ...,\n",
              "       [1.327e+01, 4.280e+00, 2.260e+00, ..., 5.900e-01, 1.560e+00,\n",
              "        8.350e+02],\n",
              "       [1.317e+01, 2.590e+00, 2.370e+00, ..., 6.000e-01, 1.620e+00,\n",
              "        8.400e+02],\n",
              "       [1.413e+01, 4.100e+00, 2.740e+00, ..., 6.100e-01, 1.600e+00,\n",
              "        5.600e+02]])"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "text/plain": [
              "array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
              "       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
              "       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1,\n",
              "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
              "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
              "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2,\n",
              "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
              "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
              "       2, 2])"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# 4) A chave \"data\" são as features e a chave \"target\" é o y. Separe os dados em 2 variáveis diferentes\n",
        "# 5) Usar todo o conjunto de dados para treinamento\n",
        "X, y = df_wine[\"data\"], df_wine[\"target\"]\n",
        "display(X)\n",
        "display(y)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 126,
      "metadata": {},
      "outputs": [],
      "source": [
        "# 6) Treinar o DBScan. Quantos clusters foram gerados com os parâmetros que vc escolheu?\n",
        "from sklearn.cluster import DBSCAN\n",
        "import numpy as np\n",
        "from sklearn import metrics"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 127,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/html": [
              "<style>#sk-container-id-15 {color: black;}#sk-container-id-15 pre{padding: 0;}#sk-container-id-15 div.sk-toggleable {background-color: white;}#sk-container-id-15 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-15 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-15 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-15 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-15 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-15 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-15 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-15 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-15 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-15 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-15 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-15 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-15 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-15 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-15 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-15 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-15 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-15 div.sk-item {position: relative;z-index: 1;}#sk-container-id-15 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-15 div.sk-item::before, #sk-container-id-15 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-15 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-15 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-15 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-15 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-15 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-15 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-15 div.sk-label-container {text-align: center;}#sk-container-id-15 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-15 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-15\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>DBSCAN(eps=50, min_samples=20)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-17\" type=\"checkbox\" checked><label for=\"sk-estimator-id-17\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">DBSCAN</label><div class=\"sk-toggleable__content\"><pre>DBSCAN(eps=50, min_samples=20)</pre></div></div></div></div></div>"
            ],
            "text/plain": [
              "DBSCAN(eps=50, min_samples=20)"
            ]
          },
          "execution_count": 127,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "model = DBSCAN(eps=50, min_samples=20)\n",
        "model.fit(X)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 128,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/plain": [
              "array([-1, -1, -1, -1,  0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,\n",
              "       -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,\n",
              "       -1, -1, -1, -1, -1, -1, -1, -1, -1,  0, -1, -1, -1, -1, -1, -1, -1,\n",
              "       -1, -1, -1, -1, -1, -1, -1, -1,  0,  0,  0,  0,  0,  0,  0,  0,  0,\n",
              "        0,  0, -1,  0,  0, -1, -1,  0,  0,  0, -1,  0, -1,  0,  0,  0,  0,\n",
              "        0,  0,  0,  0,  0,  0,  0,  0, -1,  0, -1,  0,  0,  0,  0,  0,  0,\n",
              "        0,  0,  0, -1,  0,  0, -1,  0,  0, -1,  0,  0,  0,  0,  0,  0,  0,\n",
              "        0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,\n",
              "        0,  0,  0,  0,  0, -1,  0,  0, -1, -1,  0,  0,  0,  0,  0,  0,  0,\n",
              "        0,  0,  0,  0, -1,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,\n",
              "        0,  0,  0,  0,  0, -1, -1,  0])"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "core_samples_mask = np.zeros_like(model.labels_, dtype=bool)\n",
        "core_samples_mask[model.core_sample_indices_] = True\n",
        "labels = model.labels_\n",
        "display(labels)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 129,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/plain": [
              "array([-1,  0])"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "display(np.unique(labels))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 130,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/plain": [
              "1"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "text/plain": [
              "73"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "text/plain": [
              "'Número estimado de clusters: 1'"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "text/plain": [
              "'Número estimado de ruídos: 73'"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "n_cluters = len(set(labels)) - (1 if -1 in labels else 0)\n",
        "display(n_cluters)\n",
        "\n",
        "n_noise = list(labels).count(-1)\n",
        "display(n_noise)\n",
        "\n",
        "display(\"Número estimado de clusters: %d\" % n_cluters)\n",
        "display(\"Número estimado de ruídos: %d\" % n_noise)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 131,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/plain": [
              "'Índice de Silhouette: 0.559'"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# 7) Calcular o Coeficiente de Silhouette para o DBScan treinado em (6)\n",
        "display(\"Índice de Silhouette: %0.3f\" % metrics.silhouette_score(X, labels))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 136,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "{'eps': 51.05263157894737, 'min_samples': 5}\n"
          ]
        }
      ],
      "source": [
        "# 8) É possível melhorar o Coeficiente de Silhouette? Altere os parâmetros do DBScan para conseguir um valor melhor do coeficiente\n",
        "from sklearn.metrics import make_scorer\n",
        "\n",
        "param_grid = {'eps': np.linspace(10, 70, 20), 'min_samples': np.arange(2, 30)}\n",
        "def cmp(y_pred, y_true):\n",
        "    return np.sum(y_pred == y_true)\n",
        "\n",
        "class DBSCANWrapper(DBSCAN):\n",
        "    # Won't work if `_X` is not the same X used in `self.fit`\n",
        "    def predict(self, _X, _y=None):\n",
        "        return self.labels_\n",
        "    \n",
        "cv = [(np.arange(X.shape[0]), np.arange(X.shape[0]))]\n",
        "\n",
        "search = GridSearchCV(DBSCANWrapper(), param_grid, scoring=make_scorer(cmp), cv=cv)\n",
        "search.fit(X, y)\n",
        "print(search.best_params_)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 139,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/plain": [
              "'Índice de Silhouette: 0.490'"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "best_model = grid.best_estimator_\n",
        "best_model.fit(X)\n",
        "labels = best_model.labels_\n",
        "display(\"Índice de Silhouette: %0.3f\" % metrics.silhouette_score(X, labels))"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.10.12"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}