5_Customer_segmentation.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "B1.5.5 Customer segmentation - teacher's sheet.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lVpq08B7v485"
      },
      "source": [
        "#5) Classification of customers\n",
        "\n",
        "In this section we try to classify customers to the above clusters using some features 'mean', 'categ_0', 'categ_1', 'categ_2', 'categ_3', 'categ_4'. Thus, we can use our models to classify a customer at the first purchase.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "F7KR6Jeoxcg_"
      },
      "source": [
        "import pandas as pd\n",
        "from google.colab import drive\n",
        "drive.mount('/content/drive')\n",
        "relative_path = '/content/drive/My Drive/Course v2/B1. Machine Learning Deep Learning/B1.5 Customer Segmentation/Data/'\n",
        "selected_customers = pd.read_csv(relative_path + 'data_selected_customers.csv')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8BH59QssbwKi"
      },
      "source": [
        "import warnings\n",
        "warnings.filterwarnings('ignore') "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xgrHhluZwAwZ"
      },
      "source": [
        "We first build a class `Class_Fit` to manage classifier."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LbhsvHh3vvp5"
      },
      "source": [
        "class Class_Fit:\n",
        "    def __init__(self, clf, params=None):\n",
        "        if params:            \n",
        "            self.clf = clf(**params)\n",
        "        else:\n",
        "            self.clf = clf()\n",
        "\n",
        "    def train(self, x_train, y_train):\n",
        "        self.clf.fit(x_train, y_train)\n",
        "\n",
        "    def predict(self, x):\n",
        "        return self.clf.predict(x)\n",
        "    \n",
        "    def grid_search(self, parameters, Kfold):\n",
        "        self.grid = GridSearchCV(estimator = self.clf, param_grid = parameters, cv = Kfold)\n",
        "        \n",
        "    def grid_fit(self, X, Y):\n",
        "        self.grid.fit(X, Y)\n",
        "        \n",
        "    def grid_predict(self, X, Y):\n",
        "        self.predictions = self.grid.predict(X)\n",
        "        print(\"Precision: {:.2f} % \".format(100*metrics.accuracy_score(Y, self.predictions)))\n",
        "        "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Nd5uL1D9wSjW"
      },
      "source": [
        "##5.1) Prepare train and validation data "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Mu0g7vl2wEL6"
      },
      "source": [
        "columns = ['mean', 'categ_0', 'categ_1', 'categ_2', 'categ_3', 'categ_4' ]\n",
        "X = selected_customers[columns]\n",
        "Y = selected_customers['cluster']"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HRBF6jtmxo1t"
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "X_train, X_test, Y_train, Y_test = train_test_split(X, Y, train_size = 0.8)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NpiODyIRxu2-",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 121
        },
        "outputId": "1e1245f3-2738-41bb-d253-03dc37a06100"
      },
      "source": [
        "Y_train.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "1951    7\n",
              "3255    0\n",
              "4127    8\n",
              "978     8\n",
              "2414    8\n",
              "Name: cluster, dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hfAkt5y9xyZc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 202
        },
        "outputId": "9358eb91-2f70-4eb6-e214-949f1f7dd2c9"
      },
      "source": [
        "X_train.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>mean</th>\n",
              "      <th>categ_0</th>\n",
              "      <th>categ_1</th>\n",
              "      <th>categ_2</th>\n",
              "      <th>categ_3</th>\n",
              "      <th>categ_4</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>1951</th>\n",
              "      <td>1488.033333</td>\n",
              "      <td>0.534262</td>\n",
              "      <td>86.729688</td>\n",
              "      <td>12.736050</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3255</th>\n",
              "      <td>243.840000</td>\n",
              "      <td>6.971785</td>\n",
              "      <td>24.606299</td>\n",
              "      <td>21.817585</td>\n",
              "      <td>20.767717</td>\n",
              "      <td>25.836614</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4127</th>\n",
              "      <td>174.800000</td>\n",
              "      <td>64.759725</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>26.887872</td>\n",
              "      <td>4.061785</td>\n",
              "      <td>4.290618</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>978</th>\n",
              "      <td>124.260000</td>\n",
              "      <td>43.457267</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>32.641236</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>23.901497</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2414</th>\n",
              "      <td>218.566667</td>\n",
              "      <td>45.058716</td>\n",
              "      <td>2.333384</td>\n",
              "      <td>19.063596</td>\n",
              "      <td>4.209242</td>\n",
              "      <td>29.335062</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "             mean    categ_0    categ_1    categ_2    categ_3    categ_4\n",
              "1951  1488.033333   0.534262  86.729688  12.736050   0.000000   0.000000\n",
              "3255   243.840000   6.971785  24.606299  21.817585  20.767717  25.836614\n",
              "4127   174.800000  64.759725   0.000000  26.887872   4.061785   4.290618\n",
              "978    124.260000  43.457267   0.000000  32.641236   0.000000  23.901497\n",
              "2414   218.566667  45.058716   2.333384  19.063596   4.209242  29.335062"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xcbplt4rx_ZY"
      },
      "source": [
        "## 5.2) SVM classifier\n",
        "\n",
        "Here we build an SVM classifier. Use GridSearchCV with logariphmic scale from 0.01 to 100. For Cross-Validation use 5 folds."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4OLZOm-Dx34k"
      },
      "source": [
        "import numpy as np\n",
        "from sklearn import svm\n",
        "from sklearn.model_selection import GridSearchCV\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "svc = Class_Fit(clf = svm.LinearSVC, params={'max_iter':100})\n",
        "svc.grid_search(parameters = [{'C':np.logspace(-2,2,20)}], Kfold = 5)\n",
        "# to do - fix number of iterations"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yIvSh561yET4"
      },
      "source": [
        "from sklearn import metrics\n",
        "svc.grid_fit(X = X_train, Y = Y_train)\n",
        "#output is too big, try to supress it"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TGpvyvB4yL1E",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "84bd4e1e-d3a8-44e9-90fe-cbddb1dfb5ec"
      },
      "source": [
        " svc.grid_predict(X_test, Y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Precision: 59.24 % \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7AyS2mhRygH5"
      },
      "source": [
        "# function to plot learning curves\n",
        "def plot_learning_curve(estimator, title, X, y, ylim=None, cv=None,\n",
        "                        n_jobs=-1, train_sizes=np.linspace(.1, 1.0, 10)):\n",
        "    \"\"\"Generate a simple plot of the test and training learning curve\"\"\"\n",
        "    plt.figure()\n",
        "    plt.title(title)\n",
        "    if ylim is not None:\n",
        "        plt.ylim(*ylim)\n",
        "    plt.xlabel(\"Training examples\")\n",
        "    plt.ylabel(\"Score\")\n",
        "    train_sizes, train_scores, test_scores = learning_curve(\n",
        "        estimator, X, y, cv=cv, n_jobs=n_jobs, train_sizes=train_sizes)\n",
        "    train_scores_mean = np.mean(train_scores, axis=1)\n",
        "    train_scores_std = np.std(train_scores, axis=1)\n",
        "    test_scores_mean = np.mean(test_scores, axis=1)\n",
        "    test_scores_std = np.std(test_scores, axis=1)\n",
        "    plt.grid()\n",
        "\n",
        "    plt.fill_between(train_sizes, train_scores_mean - train_scores_std,\n",
        "                     train_scores_mean + train_scores_std, alpha=0.1, color=\"r\")\n",
        "    plt.fill_between(train_sizes, test_scores_mean - test_scores_std,\n",
        "                     test_scores_mean + test_scores_std, alpha=0.1, color=\"g\")\n",
        "    plt.plot(train_sizes, train_scores_mean, 'o-', color=\"r\", label=\"Training score\")\n",
        "    plt.plot(train_sizes, test_scores_mean, 'o-', color=\"g\", label=\"Cross-validation score\")\n",
        "\n",
        "    plt.legend(loc=\"best\")\n",
        "    return plt"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LyOVz9IvyuF-",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "7d532156-2109-402b-f4fc-5f55a075141d"
      },
      "source": [
        "from sklearn.model_selection import learning_curve\n",
        "from sklearn.model_selection import GridSearchCV\n",
        "\n",
        "# plot learning curves\n",
        "g = plot_learning_curve(svc.grid.best_estimator_,\n",
        "                        \"SVC learning curves\", X_train, Y_train, ylim = [0.6, 1.01],\n",
        "                        cv = 5,  train_sizes = [0.05, 0.1, 0.2, 0.3, 0.4, 0.5,\n",
        "                                                0.6, 0.7, 0.8, 0.9, 1])\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "h14hOniP_L5N"
      },
      "source": [
        "## 5.3) Logistic regression\n",
        "\n",
        "**Task:** Similarly, build a logistic regression using `Class_Fit`, use grid search to train. Then compute the test precision and plot learning curves."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KTMYWHJV_OvK"
      },
      "source": [
        "from sklearn.linear_model import LogisticRegression\n",
        "lr = Class_Fit(clf = LogisticRegression, params={'solver':'lbfgs', 'multi_class':'auto', 'max_iter':300})\n",
        "lr.grid_search(parameters={'C':np.logspace(-2,2,10)}, Kfold=5)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fgxRHyGr_VAj"
      },
      "source": [
        "lr.grid_fit(X_train, Y_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0wpV7aiH_Xse",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "ee86b9c6-ca06-4a02-ec33-853e9205ba9d"
      },
      "source": [
        "lr.grid.best_params_ "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'C': 0.027825594022071243}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KnJLGxY8DqR9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "34e36038-e991-4f92-94ad-fe9d41332ec3"
      },
      "source": [
        "lr.grid_predict(X_test, Y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Precision: 89.49 % \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8e2QxCs4Dvhv",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 104
        },
        "outputId": "271d112d-844d-4e58-e9d9-7b175e0648a6"
      },
      "source": [
        "lr.grid.best_estimator_"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "LogisticRegression(C=0.027825594022071243, class_weight=None, dual=False,\n",
              "                   fit_intercept=True, intercept_scaling=1, l1_ratio=None,\n",
              "                   max_iter=300, multi_class='auto', n_jobs=None, penalty='l2',\n",
              "                   random_state=None, solver='lbfgs', tol=0.0001, verbose=0,\n",
              "                   warm_start=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qIRaT-Mr6ci0"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6YdS_pF9D0Uh",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "c8004629-3735-4868-8fe5-4b8bffdf17d0"
      },
      "source": [
        "g2 = plot_learning_curve(lr.grid.best_estimator_,\n",
        "                        \"LR learning curves\", X_train, Y_train, ylim = [0.6, 1.01],\n",
        "                         cv = 5,  train_sizes = [0.05, 0.1, 0.2, 0.3, 0.4, 0.5,\n",
        "                                                 0.6, 0.7, 0.8, 0.9, 1])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ID8qc8CGgaov"
      },
      "source": [
        "## 5.4) k-nearest neighbors\n",
        "\n",
        "**Task:** Build a knn using `Class_Fit`, use grid search to train. Then compute the test precision and plot learning curves.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b20jQO7oU_CE"
      },
      "source": [
        "from sklearn.neighbors import KNeighborsClassifier\n",
        "\n",
        "knn = Class_Fit(clf = KNeighborsClassifier, params=None)\n",
        "knn.grid_search(parameters={'n_neighbors':np.concatenate((np.arange(1,10,1),\n",
        "                                                          np.arange(10,40,5),\n",
        "                                                          np.arange(40,80,10)))},\n",
        "                Kfold=5)\n",
        "knn.grid_fit(X_train, Y_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "w9vPclI0vvAA"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CAP8EF2Fgfd3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "6f6719e9-4931-46a4-d9dc-2604876a4e88"
      },
      "source": [
        "knn.grid_predict(X_test, Y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Precision: 82.56 % \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "K5N2B1fAh5Sj",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "58d810d6-7cd0-4e81-b251-ebaab15e4c34"
      },
      "source": [
        "knn.grid.best_params_"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'n_neighbors': 6}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7KmSQjKPh7tL",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 313
        },
        "outputId": "76342fbb-5f55-4a11-c858-a665d0beec07"
      },
      "source": [
        "plot_learning_curve(knn.grid.best_estimator_, 'KNN learning curves', X_train, Y_train, ylim=[0.6, 1.01],\n",
        "                   train_sizes=[0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<module 'matplotlib.pyplot' from '/usr/local/lib/python3.6/dist-packages/matplotlib/pyplot.py'>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 22
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/T+rcGSuIQJekAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "22FCzOyQiFKJ"
      },
      "source": [
        "## 5.5) Decision tree\n",
        "\n",
        "**Task:** Build a decision tree using `Class_Fit`, use grid search to train. Then compute the test precision and plot learning curves."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GSElju8sh-pK"
      },
      "source": [
        "from sklearn.tree import DecisionTreeClassifier\n",
        "\n",
        "dt = Class_Fit(clf=DecisionTreeClassifier, params=None)\n",
        "dt.grid_search(parameters={'min_samples_leaf':np.concatenate((np.arange(1,5,1),\n",
        "                                                              np.arange(5,25,5),\n",
        "                                                              np.arange(25,55,10),\n",
        "                                                              np.arange(55,200,20)))},\n",
        "               Kfold=5)\n",
        "dt.grid_fit(X_train, Y_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FyUSI0ePxluI",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "6eca6079-795c-45b1-8fb1-b9d939325b31"
      },
      "source": [
        "np.concatenate((np.arange(1,5,1),\n",
        "                                                              np.arange(5,25,5),\n",
        "                                                              np.arange(25,55,10),\n",
        "                                                              np.arange(55,200,20)))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([  1,   2,   3,   4,   5,  10,  15,  20,  25,  35,  45,  55,  75,\n",
              "        95, 115, 135, 155, 175, 195])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qDI5tFKmiJpl",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 69
        },
        "outputId": "4402ac8e-f1ea-44c1-9668-85e13fde9693"
      },
      "source": [
        "print(dt.grid.best_params_)\n",
        "print(dt.grid_predict(X_test, Y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "{'min_samples_leaf': 15}\n",
            "Precision: 89.49 % \n",
            "None\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hQFQ2oCDiR3Q",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 313
        },
        "outputId": "fe104aa4-137c-45a2-8544-36bbcea24eb9"
      },
      "source": [
        "plot_learning_curve(dt.grid.best_estimator_, 'Decision tree learning curves', X_train, Y_train, ylim=[0.6, 1.01],\n",
        "                   train_sizes=[0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<module 'matplotlib.pyplot' from '/usr/local/lib/python3.6/dist-packages/matplotlib/pyplot.py'>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 26
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\noAZ9CGStDAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "F1m_rOZiibVd"
      },
      "source": [
        "## 5.6) Random forest\n",
        "\n",
        "**Task:** Build a random forest using `Class_Fit`, use grid search to train. Then compute the test precision and plot learning curves."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CB9z0_s4iWpa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "e8ac63ba-a2e2-4916-dff0-e43661bc0669"
      },
      "source": [
        "from sklearn import metrics\n",
        "from sklearn.ensemble import RandomForestClassifier\n",
        "\n",
        "rf = Class_Fit(clf=RandomForestClassifier, params=None)\n",
        "rf.grid_search(parameters={'n_estimators':[100, 200],\n",
        "                            'max_leaf_nodes':np.arange(5,20,2)},\n",
        "                Kfold=5)\n",
        "rf.grid_fit(X_train, Y_train)\n",
        "\n",
        "rf.grid.best_params_, rf.grid_predict(X_test, Y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Precision: 89.15 % \n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "({'max_leaf_nodes': 17, 'n_estimators': 200}, None)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 27
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6CzVxcqVigow",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 313
        },
        "outputId": "6574e3dd-2c52-4641-854a-dcedf9a048d8"
      },
      "source": [
        "plot_learning_curve(rf.grid.best_estimator_, 'Random forest learning curves', X_train, Y_train, ylim=[0.6, 1.01],\n",
        "                   train_sizes=[0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<module 'matplotlib.pyplot' from '/usr/local/lib/python3.6/dist-packages/matplotlib/pyplot.py'>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nEGZmVsoJwszMSv0G7n9cXkBdUuwAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dsptM0u3ixv8"
      },
      "source": [
        "## 5.7) Adaboost\n",
        "\n",
        "### Boosting\n",
        "\n",
        "Boosting technique was originally designed for classificaiton. The idea is to have a procedure that combines the outputs of many \"weak\" classifiers to produce a powerful \"committee\".\n",
        "\n",
        "### AdaBoost (Adaptive boosting)\n",
        "\n",
        "Let $X$ be a vector of predictors. The outcome is $Y \\in \\{-1,+1\\}$. The classifier $G(X) \\in \\{-1,+1\\}$\n",
        "\n",
        "The (training) error rate is:\n",
        "$$\\overline{err} = \\frac{1}{N} \\sum_{i=1}^N I(y_i \\neq G(x_i))$$\n",
        "\n",
        "A weak classifier has its error rate slightly better than random guessing.\n",
        "\n",
        "The purpose is to sequentially apply the weak classifier algorithm to repeatedly modified versions of thedata, thereby producing a sequence of weak classifiers $G_m(x), m=1,.. ,N$.\n",
        "\n",
        "The final prediction is a weighted vote:\n",
        "$$G(x) = sign(\\sum_{m=1}^M \\alpha_m G_m(x))$$\n",
        "$\\alpha_1,..,\\alpha_m$ are computed by the boosting algorithm. They give more influence to the more accurate classifers.\n",
        "\n",
        "Data modifications at each boosting step consists of applyting weights $w_1,..,w_N$ to each training observations $(x_i,y_i),i=1,..,N$. All weights are initially set to $\\frac{1}{N}$. For each step, the weights of misclassified observations increase, so that new classifiers pay attention to guess them correctly.\n",
        "\n",
        "---\n",
        "\n",
        "Alorithm AdaBoost.M1 (discrete AdaBoost)\n",
        "\n",
        "---\n",
        "1. Init: $w_i=\\frac{1}{N}, i=1,..,N$\n",
        "2. For m=1,..,M:\n",
        "  \n",
        "  a. Fit a classifier $G_m(x)$ to the training data using weights $w_i$\n",
        "  \n",
        "  b. Compute \n",
        "  $$\\overline{err} = \\frac{\\sum_{i=1}^N w_i I(y_i \\neq G(x_i))}{\\sum_{i=1}^N w_i}$$\n",
        "  \n",
        "  c. Compute\n",
        "  $$\\alpha_m = log(\\frac{1-\\overline{err}}{\\overline{err}})$$\n",
        "  \n",
        "  d. Update $w_i = w_i \\exp(\\alpha_m I(y_i \\neq G(x_i))), i=1,..,N$\n",
        "  \n",
        "3. Output $G(x) = sign(\\sum_{m=1}^M \\alpha_m G_m(x))$\n",
        "  \n",
        "---\n",
        "\n",
        "We often use a tree as weak learner, for example a stump.\n",
        "\n",
        "Five years later, it was discovered that AdaBoost can be seen as Forward Stagewise Additive Modeling with exponential loss $L(y, f(x))=\\exp(-yf(x))$\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SnTWu3Dii5Oa"
      },
      "source": [
        "___\n",
        "Forwad Stgewise Additive Modeling\n",
        "___\n",
        "1. Init: $f_0(x) = 0$\n",
        "2. For $m=1,..,M$:\n",
        "  \n",
        "  a. Compute: \n",
        "  $$(\\beta_m, \\gamma_m) = \\arg\\min_{\\beta, \\gamma} \\sum_{i=1}^N L(y_i, f_{m-1}(x_i) + \\beta b(x_i; \\gamma))$$\n",
        "  \n",
        "  b. Update $$f_m(x) = f_{m-1}(x) + \\beta_m b(x_i; \\gamma_m))$$\n",
        "___\n",
        "\n",
        "Here, $b(x; \\gamma)$ is a function with parameter set $\\gamma$; $\\beta_m$ is the expansion coefficient at step m. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1oP4rilUi-Cd"
      },
      "source": [
        "**Task:** Build an AdaBoost model using `Class_Fit`, use grid search to train. Then compute the test precision and plot learning curves."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8D_JlVVFiqxv",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 191
        },
        "outputId": "6b29be2a-4568-4003-e94f-f3d3b898c597"
      },
      "source": [
        "from sklearn.ensemble import AdaBoostClassifier\n",
        "\n",
        "ada = Class_Fit(clf=AdaBoostClassifier, params=None)\n",
        "\n",
        "para_grid = {'base_estimator':[DecisionTreeClassifier(max_depth=1),\n",
        "                               DecisionTreeClassifier(max_depth=5),\n",
        "                               DecisionTreeClassifier(max_depth=7),\n",
        "                               DecisionTreeClassifier(max_depth=9)\n",
        "                              ],\n",
        "             'algorithm':['SAMME','SAMME.R'],\n",
        "             'n_estimators':[200] \n",
        "            }\n",
        "ada.grid_search(parameters=para_grid, Kfold=5)\n",
        "\n",
        "ada.grid_fit(X_train, Y_train)\n",
        "\n",
        "ada.grid.best_params_, ada.grid_predict(X_test, Y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Precision: 91.80 % \n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "({'algorithm': 'SAMME',\n",
              "  'base_estimator': DecisionTreeClassifier(ccp_alpha=0.0, class_weight=None, criterion='gini',\n",
              "                         max_depth=9, max_features=None, max_leaf_nodes=None,\n",
              "                         min_impurity_decrease=0.0, min_impurity_split=None,\n",
              "                         min_samples_leaf=1, min_samples_split=2,\n",
              "                         min_weight_fraction_leaf=0.0, presort='deprecated',\n",
              "                         random_state=None, splitter='best'),\n",
              "  'n_estimators': 200},\n",
              " None)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "I7IS1xJbjAcC",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 313
        },
        "outputId": "6302cd70-900d-423c-a11d-7966dfb5ffa8"
      },
      "source": [
        "plot_learning_curve(ada.grid.best_estimator_, 'Adaboost learning curves', X_train, Y_train, ylim=[0.6, 1.01],\n",
        "                   train_sizes=[0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<module 'matplotlib.pyplot' from '/usr/local/lib/python3.6/dist-packages/matplotlib/pyplot.py'>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 30
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nzmXyBOGccy6TJwjnnHOZfgHy728J0b3Y0wAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sRyLAky5jyBp"
      },
      "source": [
        "## 5.8) Gradient boosting classifier\n",
        "\n",
        "In gradient boosting we use a differentiable loss function $L(y, f(x))$, not exponential loss like AdaBoost. We might say that Gradient boosting is more general.\n",
        "\n",
        "For each step, to find the new base learner, Gradient Boosting uses functional gradient descent. \n",
        "\n",
        "A functional is a function of functions. It maps a function to a real number, e.g. maps a hypothesis function to the loss.\n",
        "\n",
        "$$R[f]=\\sum_{n=1}^N L(y_i, f(x_i))$$\n",
        "\n",
        "Other examples:\n",
        "\n",
        "$$R[f]=f(x)$$\n",
        "$$R[f]=||f||$$\n",
        "$$R[f]=||f||^2$$\n",
        "\n",
        "\n",
        "The idea is that the functional gradient is a function, which corresponds to the stepest descent direct of the functional at the current function (similar to gradient). It's defined as follows:\n",
        "\n",
        "$$R[f + \\epsilon g] = R[f] + \\epsilon <\\nabla R[f], g> + O(\\epsilon^2) \\\\\n",
        "\\lim_{\\epsilon \\rightarrow 0} \\frac{R[f + \\epsilon g] - R[f]}{\\epsilon} = <\\nabla R[f], g>$$\n",
        "\n",
        "where $\\nabla R[f]$ is the functional gradient at $f$.\n",
        "\n",
        "Chain rule for functional gradients:\n",
        "\n",
        "$$\\nabla C(R[f]) = C' (R[f]) \\nabla R[f]$$\n",
        "where C is a function.\n",
        "\n",
        "Examples:\n",
        "\n",
        "$$\\nabla (||f||^2)=2f$$\n",
        "$$\\nabla ((||f||^2)^3)=3||f||^4 2f = 6||f||^4 f$$\n",
        "\n",
        "Note: the definition of functional gradient above depends on the choice of the inner product.\n",
        "\n",
        "With N observations, we use a approximate inner product:\n",
        "$$<f,g> = \\frac{1}{N} \\sum_{n=1}^N f(x_n) g(x_n)$$\n",
        "\n",
        "When we have the functional gradient, we apply its opposite on $\\{x_i\\}$ to obtain pseudo-residuals:\n",
        "\n",
        "$$r_{im}=-[ \\frac{\\partial L (y_i, f(x_i)) }{\\partial f(x_i)}]_{f=f_{m-1}}$$\n",
        "\n",
        "Examples of gradient descent (residuals):\n",
        "1. $$L(y_i, f(x_i)) = \\frac{1}{2} (y_i- f(x_i))^2\\\\\n",
        "-\\frac{\\partial L (y_i, f(x_i)) }{\\partial f(x_i)}=y_i - f(x_i)$$\n",
        "\n",
        "2. $$L(y_i, f(x_i)) =  |y_i- f(x_i)|\\\\\n",
        "-\\frac{\\partial L (y_i, f(x_i)) }{\\partial f(x_i)}=sign(y_i - f(x_i))$$\n",
        "\n",
        "---\n",
        "\n",
        "Gradient Tree Boosting Algorithm (regresion, classification is similar)\n",
        "\n",
        "---\n",
        "1. Init: $f_0(x) = \\arg\\min_{\\gamma} \\sum_{i=1}^N L(y_i, \\gamma)$\n",
        "2. For $m=1,..,M$:\n",
        "  \n",
        "  a. For $i=1,..,N$, compute the pseudo-residuals:\n",
        "  $$r_{im}=-[ \\frac{\\partial L (y_i, f(x_i)) }{\\partial f(x_i)}]_{f=f_{m-1}}$$\n",
        "  \n",
        "  b. Fit a regression tree to the targets $r_{im}$ (i.e. data $\\{x_i, r_{im}\\}_{i=1}^N$) giving terminal regions $R_{jm}, j= 1,2,...,J_m$.\n",
        "  \n",
        "  c. For $j= 1,2,...,J_m$ compute:\n",
        "  $$\\gamma_{jm} = \\arg\\min_{\\gamma} \\sum_{x_i \\in R_{jm}}^N L(y_i, f(x_i) + \\gamma)$$\n",
        "  \n",
        "  d. Update $$f_m(x) = f_{m-1}(x) + \\sum_{j=1}^{J_m} \\gamma_{jm} I(x \\in R_{jm})$$\n",
        "  \n",
        "3. Output $f_M(x)$\n",
        "  \n",
        "---\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TvE4EGDQkFJG"
      },
      "source": [
        "**Task:** Build a gradient boosting model using `Class_Fit`, use grid search to train. Then compute the test precision and plot learning curves."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "O1iqlZU4js-X",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "13132158-53c0-482b-f037-9c117f48ee23"
      },
      "source": [
        "from sklearn.ensemble import GradientBoostingClassifier \n",
        "gb = Class_Fit(clf=GradientBoostingClassifier, params=None)\n",
        "\n",
        "para_grid = {\n",
        "    'n_estimators':[100, 200, 300]\n",
        "            }\n",
        "gb.grid_search(parameters=para_grid, Kfold=5)\n",
        "\n",
        "gb.grid_fit(X_train, Y_train)\n",
        "\n",
        "gb.grid.best_params_, ada.grid_predict(X_test, Y_test)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Precision: 91.80 % \n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "({'n_estimators': 100}, None)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 31
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iaasUcy0kHoK",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 313
        },
        "outputId": "f863f32b-3e5f-4a7e-c007-f4cce4b36e74"
      },
      "source": [
        "plot_learning_curve(gb.grid.best_estimator_, 'Gradient boosting learning curves', X_train, Y_train, ylim=[0.6, 1.01],\n",
        "                   train_sizes=[0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<module 'matplotlib.pyplot' from '/usr/local/lib/python3.6/dist-packages/matplotlib/pyplot.py'>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 32
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bLtqUWg7kMng"
      },
      "source": [
        "## 5.9) XGBoost \n",
        "\n",
        "XGBoost is an optimized distributed gradient boosting library designed to be highly efficient, flexible and portable.\n",
        "\n",
        "https://xgboost.readthedocs.io/en/latest/"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PSVpSwmukLsG",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "ef82a04d-faf1-44df-b007-5324c36ee0e3"
      },
      "source": [
        "from xgboost import XGBClassifier\n",
        "xgb = Class_Fit(clf=XGBClassifier, params=None)\n",
        "\n",
        "para_grid = {\n",
        "    'objective':['multi:softmax'],\n",
        "    'n_estimators':[100, 200, 300]\n",
        "            }\n",
        "xgb.grid_search(parameters=para_grid, Kfold=5)\n",
        "\n",
        "xgb.grid_fit(X_train, Y_train)\n",
        "\n",
        "xgb.grid.best_params_, xgb.grid_predict(X_test, Y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Precision: 91.80 % \n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "({'n_estimators': 100, 'objective': 'multi:softmax'}, None)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 33
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Zw4MN2_akVXN",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 313
        },
        "outputId": "272e2a17-531e-46e3-c95f-1b24175a96a4"
      },
      "source": [
        "plot_learning_curve(xgb.grid.best_estimator_, 'Gradient boosting learning curves', X_train, Y_train, ylim=[0.6, 1.01],\n",
        "                   train_sizes=[0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<module 'matplotlib.pyplot' from '/usr/local/lib/python3.6/dist-packages/matplotlib/pyplot.py'>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 34
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "efkFjBgNkoO8"
      },
      "source": [
        "# 5.10) Testing predictions\n",
        "\n",
        "**Task:** For each trained model above, compute the precision on the test set."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1K32Y5wirrfY"
      },
      "source": [
        "**Task:** For each trained model above, compute the precision on the test set."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PPVWtZ-BqB0_",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 381
        },
        "outputId": "2151ca40-0b5f-40bc-d9ca-431c6c98d68d"
      },
      "source": [
        "classifiers = [(svc, 'Support Vector Machine'),\n",
        "                (lr, 'Logistic Regression'),\n",
        "                (knn, 'k-Nearest Neighbors'),\n",
        "                (dt, 'Decision Tree'),\n",
        "                (rf, 'Random Forest'),\n",
        "                (gb, 'Gradient Boosting'),\n",
        "                (xgb, 'XGBoost')]\n",
        "#______________________________\n",
        "for clf, label in classifiers:\n",
        "    print(30*'_', '\\n{}'.format(label))\n",
        "    clf.grid_predict(X_test, Y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "______________________________ \n",
            "Support Vector Machine\n",
            "Precision: 58.89 % \n",
            "______________________________ \n",
            "Logistic Regression\n",
            "Precision: 91.34 % \n",
            "______________________________ \n",
            "k-Nearest Neighbors\n",
            "Precision: 82.56 % \n",
            "______________________________ \n",
            "Decision Tree\n",
            "Precision: 89.49 % \n",
            "______________________________ \n",
            "Random Forest\n",
            "Precision: 89.15 % \n",
            "______________________________ \n",
            "Gradient Boosting\n",
            "Precision: 90.88 % \n",
            "______________________________ \n",
            "XGBoost\n",
            "Precision: 91.57 % \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "55s5zQRFAqxA"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}