Churn-Prediction-in-a-Telco_colab.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Churn_Telecom_v3.0.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "3Hs1osCSdjsj"
      ],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/alonsosilvaallende/Churn-Prediction-in-a-Telco/blob/master/Churn-Prediction-in-a-Telco_colab.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uBN1uaBEmMa7",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# install lifelines and pysurvival\n",
        "pip install -q lifelines pysurvival"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SmQ7emzZhket",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# usual imports\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "from matplotlib import pyplot as plt\n",
        "\n",
        "%matplotlib inline"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0Bl87qQJf-QZ",
        "colab_type": "text"
      },
      "source": [
        "# Churn prediction in a telco using random forests"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kurG8Szkulo9",
        "colab_type": "text"
      },
      "source": [
        "Customer churn is the loss of clients or customers. In order to avoid losing customers, a company needs to examine why its customers have left in the past and which features are more important to determine who will churn next. Our task is therefore to predict whether customers are about to churn and which are the most important features to get that prediction. How can we use machine learning to better predict customer churn?\n",
        "\n",
        "Given a dataset how can you evaluate if your churn predictions were right or not? Everybody will eventually churn (some people will change to a different company, some people will stop using the services altogether, and eventually all the people will churn if not because everybody eventually dies).\n",
        "So the question remains, how do you evaluate if your churn predictions were right or wrong?\n",
        "\n",
        "It has been argued elsewhere (see for example [here](https://ragulpr.github.io/2016/12/22/WTTE-RNN-Hackless-churn-modeling/)) that the right approach to evaluate churn prediction models is through the concordance index and I'm convinced that it is a better approach than to treat it as a binary classification problem because of the reasons explained above (although you are free to try binary classification as well).\n",
        "I think the right way to think about the problem is that given the customers our model has predicted to leave in a certain order, that that order has been right. The customers that were predicted to churn first have indeed churned first. Therefore, the right appoach is to evaluate our machine learning models is through the concordance index.\n",
        "\n",
        "In this post, I will use different kinds of random forests to predict churn in a telco.\n",
        "\n",
        "Treselle Systems, a data consulting service, [analyzed customer churn data using logistic regression](http://www.treselle.com/blog/customer-churn-logistic-regression-with-r/).\n",
        "\n",
        "We will use that dataset to do our analysis. The dataset includes information about:\n",
        "\n",
        "+ Customers who left within the last month – the column is called Churn\n",
        "+ Services that each customer has signed up for – phone, multiple lines, internet, online security, online backup, device protection, tech support, and streaming TV and movies\n",
        "+ Customer account information – how long they’ve been a customer, contract, payment method, paperless billing, monthly charges, and total charges\n",
        "+ Demographic info about customers – gender, age range, and if they have partners and dependents"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sOwy8uhHhYiW",
        "colab_type": "code",
        "outputId": "c476ae47-d5a4-4c93-ba28-2fd84d1616ee",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 700
        }
      },
      "source": [
        "churn_data = pd.read_csv('https://raw.githubusercontent.com/treselle-systems/customer_churn_analysis/master/WA_Fn-UseC_-Telco-Customer-Churn.csv')\n",
        "churn_data.head().T"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>0</th>\n",
              "      <th>1</th>\n",
              "      <th>2</th>\n",
              "      <th>3</th>\n",
              "      <th>4</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>customerID</th>\n",
              "      <td>7590-VHVEG</td>\n",
              "      <td>5575-GNVDE</td>\n",
              "      <td>3668-QPYBK</td>\n",
              "      <td>7795-CFOCW</td>\n",
              "      <td>9237-HQITU</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>gender</th>\n",
              "      <td>Female</td>\n",
              "      <td>Male</td>\n",
              "      <td>Male</td>\n",
              "      <td>Male</td>\n",
              "      <td>Female</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>SeniorCitizen</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Partner</th>\n",
              "      <td>Yes</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Dependents</th>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>tenure</th>\n",
              "      <td>1</td>\n",
              "      <td>34</td>\n",
              "      <td>2</td>\n",
              "      <td>45</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>PhoneService</th>\n",
              "      <td>No</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Yes</td>\n",
              "      <td>No</td>\n",
              "      <td>Yes</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>MultipleLines</th>\n",
              "      <td>No phone service</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>No phone service</td>\n",
              "      <td>No</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>InternetService</th>\n",
              "      <td>DSL</td>\n",
              "      <td>DSL</td>\n",
              "      <td>DSL</td>\n",
              "      <td>DSL</td>\n",
              "      <td>Fiber optic</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>OnlineSecurity</th>\n",
              "      <td>No</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Yes</td>\n",
              "      <td>Yes</td>\n",
              "      <td>No</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>OnlineBackup</th>\n",
              "      <td>Yes</td>\n",
              "      <td>No</td>\n",
              "      <td>Yes</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>DeviceProtection</th>\n",
              "      <td>No</td>\n",
              "      <td>Yes</td>\n",
              "      <td>No</td>\n",
              "      <td>Yes</td>\n",
              "      <td>No</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>TechSupport</th>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>Yes</td>\n",
              "      <td>No</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>StreamingTV</th>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>StreamingMovies</th>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Contract</th>\n",
              "      <td>Month-to-month</td>\n",
              "      <td>One year</td>\n",
              "      <td>Month-to-month</td>\n",
              "      <td>One year</td>\n",
              "      <td>Month-to-month</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>PaperlessBilling</th>\n",
              "      <td>Yes</td>\n",
              "      <td>No</td>\n",
              "      <td>Yes</td>\n",
              "      <td>No</td>\n",
              "      <td>Yes</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>PaymentMethod</th>\n",
              "      <td>Electronic check</td>\n",
              "      <td>Mailed check</td>\n",
              "      <td>Mailed check</td>\n",
              "      <td>Bank transfer (automatic)</td>\n",
              "      <td>Electronic check</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>MonthlyCharges</th>\n",
              "      <td>29.85</td>\n",
              "      <td>56.95</td>\n",
              "      <td>53.85</td>\n",
              "      <td>42.3</td>\n",
              "      <td>70.7</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>TotalCharges</th>\n",
              "      <td>29.85</td>\n",
              "      <td>1889.5</td>\n",
              "      <td>108.15</td>\n",
              "      <td>1840.75</td>\n",
              "      <td>151.65</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Churn</th>\n",
              "      <td>No</td>\n",
              "      <td>No</td>\n",
              "      <td>Yes</td>\n",
              "      <td>No</td>\n",
              "      <td>Yes</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                 0  ...                 4\n",
              "customerID              7590-VHVEG  ...        9237-HQITU\n",
              "gender                      Female  ...            Female\n",
              "SeniorCitizen                    0  ...                 0\n",
              "Partner                        Yes  ...                No\n",
              "Dependents                      No  ...                No\n",
              "tenure                           1  ...                 2\n",
              "PhoneService                    No  ...               Yes\n",
              "MultipleLines     No phone service  ...                No\n",
              "InternetService                DSL  ...       Fiber optic\n",
              "OnlineSecurity                  No  ...                No\n",
              "OnlineBackup                   Yes  ...                No\n",
              "DeviceProtection                No  ...                No\n",
              "TechSupport                     No  ...                No\n",
              "StreamingTV                     No  ...                No\n",
              "StreamingMovies                 No  ...                No\n",
              "Contract            Month-to-month  ...    Month-to-month\n",
              "PaperlessBilling               Yes  ...               Yes\n",
              "PaymentMethod     Electronic check  ...  Electronic check\n",
              "MonthlyCharges               29.85  ...              70.7\n",
              "TotalCharges                 29.85  ...            151.65\n",
              "Churn                           No  ...               Yes\n",
              "\n",
              "[21 rows x 5 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SQycED_Ko3AZ",
        "colab_type": "text"
      },
      "source": [
        "## Preprocessing"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hbRttpiAAxbO",
        "colab_type": "text"
      },
      "source": [
        "We first verify that there are no NaNs and there are no duplicated records."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bwpLOaCuW7Vg",
        "colab_type": "code",
        "outputId": "9820fae7-66cf-44dd-987d-b6a4d93d7a72",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 391
        }
      },
      "source": [
        "# count how many NaN values does each feature/covariable have\n",
        "churn_data.isnull().sum()"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "customerID          0\n",
              "gender              0\n",
              "SeniorCitizen       0\n",
              "Partner             0\n",
              "Dependents          0\n",
              "tenure              0\n",
              "PhoneService        0\n",
              "MultipleLines       0\n",
              "InternetService     0\n",
              "OnlineSecurity      0\n",
              "OnlineBackup        0\n",
              "DeviceProtection    0\n",
              "TechSupport         0\n",
              "StreamingTV         0\n",
              "StreamingMovies     0\n",
              "Contract            0\n",
              "PaperlessBilling    0\n",
              "PaymentMethod       0\n",
              "MonthlyCharges      0\n",
              "TotalCharges        0\n",
              "Churn               0\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aaCefTkPqXXz",
        "colab_type": "code",
        "outputId": "f101200c-ec0b-40db-a6e6-bf731df2fe2b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# Checking for duplicates\n",
        "N_dupli = sum(churn_data.duplicated())\n",
        "print(f'The data set contains {N_dupli} duplicates')"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "The data set contains 0 duplicates\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pHc_B96fiCXB",
        "colab_type": "code",
        "outputId": "91771e1b-e69d-4214-bad2-dc25e59e3842",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "N = churn_data.shape[0]\n",
        "hasNotChurned = sum(churn_data['Churn']=='No')\n",
        "hasChurned = sum(churn_data['Churn']=='Yes')\n",
        "\n",
        "print(f'Number of samples: {N}')\n",
        "print(f'Number of columns: {churn_data.shape[1]}')\n",
        "print(f'Number of customers that have churned: {hasChurned}')\n",
        "print(f'Number of customers that have not churned yet: {hasNotChurned}')\n",
        "print(f'Percentage of customers that have churned: {100*hasChurned/len(churn_data):.0f}%')\n",
        "print(f'Percentage of customers that have not churned yet: {100*hasNotChurned/len(churn_data):.0f}%')"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Number of samples: 7043\n",
            "Number of columns: 21\n",
            "Number of customers that have churned: 1869\n",
            "Number of customers that have not churned yet: 5174\n",
            "Percentage of customers that have churned: 27%\n",
            "Percentage of customers that have not churned yet: 73%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DNjdsQChDYHm",
        "colab_type": "text"
      },
      "source": [
        "In the dataset, there are 7043 customers, of which 1869 customers have churned and 5174 customers that have not churned yet."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Fs2AZYs-A6mO",
        "colab_type": "text"
      },
      "source": [
        "We drop the ID column and the TotalCharges column (otherwise by knowing the MonthlyCharges our model can easily deduce how many months someone has been subscribed)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jiqMaNPHq5Sg",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Drop customerID column\n",
        "churn_data = churn_data.drop('customerID', axis=1)\n",
        "\n",
        "# Drop TotalCharges column: otherwise together with MonthlyCharges you can deduce how many months you have been subscribed\n",
        "churn_data = churn_data.drop('TotalCharges', axis=1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FCilsLD9pFbM",
        "colab_type": "code",
        "outputId": "6fbf981c-d37c-4978-ec9c-fd2e537538d9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 357
        }
      },
      "source": [
        "churn_data.dtypes"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "gender               object\n",
              "SeniorCitizen         int64\n",
              "Partner              object\n",
              "Dependents           object\n",
              "tenure                int64\n",
              "PhoneService         object\n",
              "MultipleLines        object\n",
              "InternetService      object\n",
              "OnlineSecurity       object\n",
              "OnlineBackup         object\n",
              "DeviceProtection     object\n",
              "TechSupport          object\n",
              "StreamingTV          object\n",
              "StreamingMovies      object\n",
              "Contract             object\n",
              "PaperlessBilling     object\n",
              "PaymentMethod        object\n",
              "MonthlyCharges      float64\n",
              "Churn                object\n",
              "dtype: object"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "o7rmNjJ0XV_Y",
        "colab_type": "text"
      },
      "source": [
        "### Convert object variables to numerical variables."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0AjeKcPCV4lH",
        "colab_type": "text"
      },
      "source": [
        "The column of Churn is an object data type, so we transform it to a numerical data type."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yZSthdnyphtu",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "churn_data['Churn'] = churn_data['Churn'].map({'No': False, 'Yes': True})"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-nP4l4R3a4ao",
        "colab_type": "text"
      },
      "source": [
        "We use [one-hot encoding](https://en.wikipedia.org/wiki/One-hot) for the categorical variables."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TBWhrpv_SX9q",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# creating one-hot columns\n",
        "categories = ['gender', 'Partner', 'Dependents', 'PhoneService', 'MultipleLines', 'InternetService', 'OnlineSecurity', 'OnlineBackup', 'DeviceProtection', 'TechSupport', 'StreamingTV', 'StreamingMovies', 'Contract', 'PaperlessBilling', 'PaymentMethod']\n",
        "churn_data = pd.get_dummies(churn_data, columns=categories, drop_first=True, dtype=float)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iiXfEJwOuZYR",
        "colab_type": "code",
        "outputId": "a4c3f1cc-09cd-481c-ea22-52b007abd71a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 258
        }
      },
      "source": [
        "churn_data.head()"
      ],
      "execution_count": 12,
      "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>SeniorCitizen</th>\n",
              "      <th>tenure</th>\n",
              "      <th>MonthlyCharges</th>\n",
              "      <th>Churn</th>\n",
              "      <th>gender_Male</th>\n",
              "      <th>Partner_Yes</th>\n",
              "      <th>Dependents_Yes</th>\n",
              "      <th>PhoneService_Yes</th>\n",
              "      <th>MultipleLines_No phone service</th>\n",
              "      <th>MultipleLines_Yes</th>\n",
              "      <th>InternetService_Fiber optic</th>\n",
              "      <th>InternetService_No</th>\n",
              "      <th>OnlineSecurity_No internet service</th>\n",
              "      <th>OnlineSecurity_Yes</th>\n",
              "      <th>OnlineBackup_No internet service</th>\n",
              "      <th>OnlineBackup_Yes</th>\n",
              "      <th>DeviceProtection_No internet service</th>\n",
              "      <th>DeviceProtection_Yes</th>\n",
              "      <th>TechSupport_No internet service</th>\n",
              "      <th>TechSupport_Yes</th>\n",
              "      <th>StreamingTV_No internet service</th>\n",
              "      <th>StreamingTV_Yes</th>\n",
              "      <th>StreamingMovies_No internet service</th>\n",
              "      <th>StreamingMovies_Yes</th>\n",
              "      <th>Contract_One year</th>\n",
              "      <th>Contract_Two year</th>\n",
              "      <th>PaperlessBilling_Yes</th>\n",
              "      <th>PaymentMethod_Credit card (automatic)</th>\n",
              "      <th>PaymentMethod_Electronic check</th>\n",
              "      <th>PaymentMethod_Mailed check</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>29.85</td>\n",
              "      <td>False</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0</td>\n",
              "      <td>34</td>\n",
              "      <td>56.95</td>\n",
              "      <td>False</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0</td>\n",
              "      <td>2</td>\n",
              "      <td>53.85</td>\n",
              "      <td>True</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0</td>\n",
              "      <td>45</td>\n",
              "      <td>42.30</td>\n",
              "      <td>False</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0</td>\n",
              "      <td>2</td>\n",
              "      <td>70.70</td>\n",
              "      <td>True</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   SeniorCitizen  ...  PaymentMethod_Mailed check\n",
              "0              0  ...                         0.0\n",
              "1              0  ...                         1.0\n",
              "2              0  ...                         1.0\n",
              "3              0  ...                         0.0\n",
              "4              0  ...                         0.0\n",
              "\n",
              "[5 rows x 30 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YrYT2Q0NBj3e",
        "colab_type": "text"
      },
      "source": [
        "We identify the time and event of interest columns."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "o6v3vTdKd8FJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# creating the time and event columns\n",
        "event_col = 'Churn'\n",
        "time_col = 'tenure'\n",
        "\n",
        "# extracting the features/covariables\n",
        "features = np.setdiff1d(churn_data.columns, [time_col, event_col]).tolist()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-sup-x2RbhDB",
        "colab_type": "code",
        "outputId": "2c47961b-77eb-44e5-eb43-fe3204b016e4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "print(f'Number of features/covariables: {len(features)}')"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Number of features/covariables: 28\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nZjRRtaue6dR",
        "colab_type": "text"
      },
      "source": [
        "## Data Analysis"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Hs07lOZqbjNl",
        "colab_type": "text"
      },
      "source": [
        "We split the dataset into training (70%) and testing (30%) datasets."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NMXud1tVvGM_",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Building training and testing sets\n",
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "index_train, index_test = train_test_split(range(N), test_size = 0.3, random_state=0)\n",
        "data_train = churn_data.loc[index_train].reset_index(drop=True)\n",
        "data_test  = churn_data.loc[index_test].reset_index(drop=True)\n",
        "\n",
        "# Creating the X, T and E inputs\n",
        "X_trn, X_val = data_train[features], data_test[features]\n",
        "T_trn, T_val = data_train[time_col], data_test[time_col]\n",
        "E_trn, E_val = data_train[event_col], data_test[event_col]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9ntalQ8yoObF",
        "colab_type": "text"
      },
      "source": [
        "## Regression problem"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gquTdEzwB0dt",
        "colab_type": "text"
      },
      "source": [
        "In this post, I want to compare the performance of three different random forests:\n",
        "\n",
        "+ Random forest classifier\n",
        "+ Random forest regressor\n",
        "+ Random survival forest\n",
        "\n",
        "We want to determine from only the set of features/covariables who will churn first. There is no easy way to include in the learning of random forest classifier or random forest regressor both the events of interest (churn) and the times people have been subscribed.\n",
        "\n",
        "Let's first consider only the case when they are using only one information\n",
        "\n",
        "+ Random forest classifier will use the churn information\n",
        "+ Random forest regressor will use the time information\n",
        "\n",
        "and then we will compare with a combination of both random forest classifier and random forest regressor:\n",
        "\n",
        "+ Random forest classifier will use the churn information and we will use that churn information to run a random forest regressor (we call that approach Random forest classifier-regressor).\n",
        "\n",
        "+ Random forest regressor will use the time information and we will use that time information to run a random forest classifier (we call that approach Random forest regressor-classifier)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_7ZjpzHBgbAS",
        "colab_type": "code",
        "outputId": "e6461dcf-ac71-42e0-9799-06fec5354f13",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        }
      },
      "source": [
        "from sklearn.ensemble import RandomForestClassifier\n",
        "from lifelines.utils import concordance_index\n",
        "\n",
        "rfc = RandomForestClassifier(n_estimators=100, random_state=0)\n",
        "rfc.fit(X_trn, E_trn)\n",
        "\n",
        "ci_rfc_trn = concordance_index(T_trn, rfc.predict_proba(X_trn)[:,0], E_trn)\n",
        "ci_rfc_val = concordance_index(T_val, rfc.predict_proba(X_val)[:,0], E_val)\n",
        "\n",
        "print(f'Concordance index of Random forest classifier: train: {ci_rfc_trn:.3f} valid: {ci_rfc_val:.3f}')\n",
        "\n",
        "# feature importance\n",
        "fi_rfc = rfc.feature_importances_\n",
        "rfi_rfc =  np.abs(fi_rfc)/np.sum(np.abs(fi_rfc))\n",
        "lab = X_trn.columns\n",
        "df_fi_rfc = pd.DataFrame(data={'importance_rfc': rfi_rfc}, index=lab)\n",
        "df_fi_rfc = df_fi_rfc.sort_values(by='importance_rfc', ascending=True)[-10:]\n",
        "df_fi_rfc.plot.barh(y='importance_rfc');"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Concordance index of Random forest classifier: train: 0.917 valid: 0.765\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAg8AAAD8CAYAAADwvEc8AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nojh5MDMzs6I4eTAzM7Oi/P8cZZ6ym3lbdwAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TpapOrKDrQ-y",
        "colab_type": "code",
        "outputId": "cfb5b480-b0b7-4c05-eb16-35ea2bd22666",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 359
        }
      },
      "source": [
        "df_fi_rfc.sort_values(by='importance_rfc', ascending=False)"
      ],
      "execution_count": 20,
      "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>importance_rfc</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>MonthlyCharges</th>\n",
              "      <td>0.346583</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Contract_Two year</th>\n",
              "      <td>0.051150</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>PaymentMethod_Electronic check</th>\n",
              "      <td>0.050140</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Contract_One year</th>\n",
              "      <td>0.046801</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>gender_Male</th>\n",
              "      <td>0.041857</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>OnlineSecurity_Yes</th>\n",
              "      <td>0.037556</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>InternetService_Fiber optic</th>\n",
              "      <td>0.036558</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>PaperlessBilling_Yes</th>\n",
              "      <td>0.036296</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>OnlineBackup_Yes</th>\n",
              "      <td>0.035264</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Partner_Yes</th>\n",
              "      <td>0.034986</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                importance_rfc\n",
              "MonthlyCharges                        0.346583\n",
              "Contract_Two year                     0.051150\n",
              "PaymentMethod_Electronic check        0.050140\n",
              "Contract_One year                     0.046801\n",
              "gender_Male                           0.041857\n",
              "OnlineSecurity_Yes                    0.037556\n",
              "InternetService_Fiber optic           0.036558\n",
              "PaperlessBilling_Yes                  0.036296\n",
              "OnlineBackup_Yes                      0.035264\n",
              "Partner_Yes                           0.034986"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 20
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bZG-iQ6WuFfD",
        "colab_type": "code",
        "outputId": "cb3f5602-b8d0-4327-ea95-6ee3f195dae8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        }
      },
      "source": [
        "from sklearn.ensemble import RandomForestRegressor\n",
        "\n",
        "rfr1 = RandomForestRegressor(n_estimators=100, random_state=42)\n",
        "\n",
        "X_trn1 = X_trn[E_trn == True]\n",
        "T_trn1 = T_trn[E_trn == True]\n",
        "E_trn1 = E_trn[E_trn == True]\n",
        "\n",
        "rfr1.fit(X_trn1, T_trn1)\n",
        "ci_rfr1_trn = concordance_index(T_trn1, rfr1.predict(X_trn1), E_trn1)\n",
        "ci_rfr1_val = concordance_index(T_val, rfr1.predict(X_val), E_val)\n",
        "\n",
        "print(f'Concordance index of Random forest regressor: train: {ci_rfr1_trn:.3f} valid: {ci_rfr1_val:.3f}')\n",
        "\n",
        "# feature importance\n",
        "fi_rfr1 = rfr1.feature_importances_\n",
        "rfi_rfr1 =  np.abs(fi_rfr1)/np.sum(np.abs(fi_rfr1))\n",
        "lab = X_trn.columns\n",
        "df_fi_rfr1 = pd.DataFrame(data={'importance_rfr1': rfi_rfr1}, index=lab)\n",
        "df_fi_rfr1 = df_fi_rfr1.sort_values(by='importance_rfr1', ascending=True)[18:]\n",
        "df_fi_rfr1.plot.barh(y='importance_rfr1');"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Concordance index of Random forest regressor: train: 0.919 valid: 0.817\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjIAAAD8CAYAAABtsU60AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nr7KklJmoAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hOV7R_s23hzb",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 359
        },
        "outputId": "a4098e62-9da9-4af3-d17b-8c17d4ff6656"
      },
      "source": [
        "df_fi_rfr1.sort_values(by='importance_rfr1', ascending=False)"
      ],
      "execution_count": 22,
      "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>importance_rfr1</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>MonthlyCharges</th>\n",
              "      <td>0.466281</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Contract_One year</th>\n",
              "      <td>0.105160</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Contract_Two year</th>\n",
              "      <td>0.079135</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>OnlineBackup_Yes</th>\n",
              "      <td>0.043223</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Partner_Yes</th>\n",
              "      <td>0.042977</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>MultipleLines_Yes</th>\n",
              "      <td>0.035529</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>SeniorCitizen</th>\n",
              "      <td>0.028447</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>gender_Male</th>\n",
              "      <td>0.027198</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>PaymentMethod_Credit card (automatic)</th>\n",
              "      <td>0.019719</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>PaymentMethod_Electronic check</th>\n",
              "      <td>0.019448</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                       importance_rfr1\n",
              "MonthlyCharges                                0.466281\n",
              "Contract_One year                             0.105160\n",
              "Contract_Two year                             0.079135\n",
              "OnlineBackup_Yes                              0.043223\n",
              "Partner_Yes                                   0.042977\n",
              "MultipleLines_Yes                             0.035529\n",
              "SeniorCitizen                                 0.028447\n",
              "gender_Male                                   0.027198\n",
              "PaymentMethod_Credit card (automatic)         0.019719\n",
              "PaymentMethod_Electronic check                0.019448"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 22
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "DTDLpI37dQ7X",
        "outputId": "274c9aad-a564-467a-c9c6-5b99630a0e3f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        }
      },
      "source": [
        "from sklearn.ensemble import RandomForestRegressor\n",
        "\n",
        "rfr = RandomForestRegressor(n_estimators=100, random_state=42)\n",
        "rfr.fit(X_trn, T_trn)\n",
        "\n",
        "ci_rfr_trn = concordance_index(T_trn, rfr.predict(X_trn), E_trn)\n",
        "ci_rfr_val = concordance_index(T_val, rfr.predict(X_val), E_val)\n",
        "\n",
        "print(f'Concordance index of Random forest regressor: train: {ci_rfr_trn:.3f} valid: {ci_rfr_val:.3f}')\n",
        "\n",
        "# feature importance\n",
        "fi_rfr = rfr.feature_importances_\n",
        "rfi_rfr =  np.abs(fi_rfr)/np.sum(np.abs(fi_rfr))\n",
        "lab = X_trn.columns\n",
        "df_fi_rfr = pd.DataFrame(data={'importance_rfr': rfi_rfr}, index=lab)\n",
        "df_fi_rfr = df_fi_rfr.sort_values(by='importance_rfr', ascending=True)[18:]\n",
        "df_fi_rfr.plot.barh(y='importance_rfr');"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Concordance index of Random forest regressor: train: 0.938 valid: 0.820\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAf8AAAD8CAYAAAB5Eq2kAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UDcu9rHOr8_6",
        "colab_type": "code",
        "outputId": "fc5e142a-dd5c-4b1c-f2b0-28e543e07f8c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 359
        }
      },
      "source": [
        "df_fi_rfr.sort_values(by='importance_rfr', ascending=False)"
      ],
      "execution_count": 24,
      "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>importance_rfr</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>Contract_Two year</th>\n",
              "      <td>0.312050</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>MonthlyCharges</th>\n",
              "      <td>0.257313</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Contract_One year</th>\n",
              "      <td>0.145683</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>PaymentMethod_Mailed check</th>\n",
              "      <td>0.034982</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Partner_Yes</th>\n",
              "      <td>0.028924</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>OnlineBackup_Yes</th>\n",
              "      <td>0.027996</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>MultipleLines_Yes</th>\n",
              "      <td>0.025800</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>gender_Male</th>\n",
              "      <td>0.022176</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>PaperlessBilling_Yes</th>\n",
              "      <td>0.019041</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Dependents_Yes</th>\n",
              "      <td>0.016708</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                            importance_rfr\n",
              "Contract_Two year                 0.312050\n",
              "MonthlyCharges                    0.257313\n",
              "Contract_One year                 0.145683\n",
              "PaymentMethod_Mailed check        0.034982\n",
              "Partner_Yes                       0.028924\n",
              "OnlineBackup_Yes                  0.027996\n",
              "MultipleLines_Yes                 0.025800\n",
              "gender_Male                       0.022176\n",
              "PaperlessBilling_Yes              0.019041\n",
              "Dependents_Yes                    0.016708"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lhCSEA5hp4t-",
        "colab_type": "code",
        "outputId": "6f363eee-a751-4015-b03c-61236e213eed",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        }
      },
      "source": [
        "from pysurvival.models.survival_forest import RandomSurvivalForestModel\n",
        "\n",
        "rsf = RandomSurvivalForestModel(num_trees=100)\n",
        "rsf.fit(X_trn, T_trn, E_trn, seed=0)\n",
        "ci_rsf_trn = concordance_index(T_trn, -rsf.predict_risk(X_trn), E_trn)\n",
        "ci_rsf_val = concordance_index(T_val, -rsf.predict_risk(X_val), E_val)\n",
        "\n",
        "print(f'Concordance index of Random survival forest: train: {ci_rsf_trn:.3f}, valid: {ci_rsf_val:.3f}')\n",
        "\n",
        "# feature importance\n",
        "fi_rsf = rsf.variable_importance_\n",
        "rfi_rsf =  np.abs(fi_rsf)/np.sum(np.abs(fi_rsf))\n",
        "lab = X_trn.columns\n",
        "df_fi_rsf = pd.DataFrame(data={'importance_rsf': rfi_rsf}, index=lab)\n",
        "df_fi_rsf = df_fi_rsf.sort_values(by='importance_rsf', ascending=True)[-10:]\n",
        "df_fi_rsf.plot.barh(y='importance_rsf');"
      ],
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Concordance index of Random survival forest: train: 0.847, valid: 0.844\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAg8AAAD8CAYAAADwvEc8AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YQBUISNKtBZS",
        "colab_type": "code",
        "outputId": "38bf98fe-fdc6-4355-9f23-dc1096b24b4e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 359
        }
      },
      "source": [
        "df_fi_rsf.sort_values(by='importance_rsf', ascending=False)"
      ],
      "execution_count": 28,
      "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>importance_rsf</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>Contract_Two year</th>\n",
              "      <td>0.103379</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Contract_One year</th>\n",
              "      <td>0.084170</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>PaymentMethod_Electronic check</th>\n",
              "      <td>0.066108</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Partner_Yes</th>\n",
              "      <td>0.065598</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>OnlineSecurity_Yes</th>\n",
              "      <td>0.063223</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>TechSupport_Yes</th>\n",
              "      <td>0.059835</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>OnlineBackup_Yes</th>\n",
              "      <td>0.055889</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>DeviceProtection_Yes</th>\n",
              "      <td>0.051305</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>MonthlyCharges</th>\n",
              "      <td>0.045479</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>InternetService_Fiber optic</th>\n",
              "      <td>0.036056</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                importance_rsf\n",
              "Contract_Two year                     0.103379\n",
              "Contract_One year                     0.084170\n",
              "PaymentMethod_Electronic check        0.066108\n",
              "Partner_Yes                           0.065598\n",
              "OnlineSecurity_Yes                    0.063223\n",
              "TechSupport_Yes                       0.059835\n",
              "OnlineBackup_Yes                      0.055889\n",
              "DeviceProtection_Yes                  0.051305\n",
              "MonthlyCharges                        0.045479\n",
              "InternetService_Fiber optic           0.036056"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "t3aJ4oVYlVCL",
        "colab_type": "text"
      },
      "source": [
        "## Random forest classifier-regressor\n",
        "\n",
        "Predict the event and then compute the order."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Oa1lXWRyv5_H",
        "colab_type": "code",
        "outputId": "6c46f29e-523e-400a-a71f-6d31d6a348c1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "rfc = RandomForestClassifier(n_estimators=100, random_state=0)\n",
        "rfc.fit(X_trn, E_trn)\n",
        "\n",
        "# add to X_val the predicted event\n",
        "X_val_plus_E = X_val.copy()\n",
        "X_val_plus_E[event_col] = rfc.predict(X_val)\n",
        "\n",
        "# for the training dataset add the event column\n",
        "feat_plus_E = np.union1d(features, [event_col])\n",
        "X_trn_plus_E = data_train[feat_plus_E]\n",
        "\n",
        "rfr = RandomForestRegressor(n_estimators=100, random_state=0)\n",
        "rfr.fit(X_trn_plus_E, T_trn)\n",
        "\n",
        "ci_rfcr_trn = concordance_index(T_trn, rfr.predict(X_trn_plus_E), E_trn)\n",
        "ci_rfcr_val = concordance_index(T_val, rfr.predict(X_val_plus_E), E_val)\n",
        "\n",
        "print(f'Concordance index of Random forest classifier-regressor: train: {ci_rfcr_trn:.3f}, valid: {ci_rfcr_val:.3f}')"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Concordance index of Random forest classifier-regressor: train: 0.957, valid: 0.689\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SExVtJazvk19",
        "colab_type": "text"
      },
      "source": [
        "## Random forest regressor-classifier\n",
        "\n",
        "Predict the time and then the probability of the event."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "09Q5f_3K0F-P",
        "colab_type": "code",
        "outputId": "d42c1a98-1ece-4f4c-8128-8aebec0442ac",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "rfr = RandomForestRegressor(n_estimators=100, random_state=0)\n",
        "rfr.fit(X_trn, T_trn)\n",
        "\n",
        "# add to X_val the predicted time\n",
        "X_val_plus_T = X_val.copy()\n",
        "X_val_plus_T[time_col] = rfr.predict(X_val)\n",
        "\n",
        "# for the training dataset add the time column\n",
        "feat_plus_T = np.union1d(features, [time_col])\n",
        "X_trn_plus_T = data_train[feat_plus_T]\n",
        "\n",
        "rfc = RandomForestClassifier(n_estimators=100, random_state=0)\n",
        "rfc.fit(X_trn_plus_T, E_trn)\n",
        "\n",
        "ci_rfrc_trn = concordance_index(T_trn, ~rfc.predict(X_trn_plus_T), E_trn)\n",
        "ci_rfrc_val = concordance_index(T_val, ~rfc.predict(X_val_plus_T), E_val)\n",
        "print(f'Concordance index of Random forest regressor-classifier: train: {ci_rfrc_trn:.3f}, valid: {ci_rfrc_val:.3f}')"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Concordance index of Random forest regressor-classifier: train: 0.904, valid: 0.622\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "I-4sIdRdrbcm",
        "colab_type": "text"
      },
      "source": [
        "## Box plots"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mNrSxQT-rfYK",
        "colab_type": "text"
      },
      "source": [
        "### Box plot classification"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0qVztbZMrjhd",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# choose 20 random seeds\n",
        "np.random.seed(0)\n",
        "seeds = np.random.permutation(1000)[:20]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vWfNbXvXBboZ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "rfc_ci = []\n",
        "rfr1_ci = []\n",
        "rfr2_ci = []\n",
        "rsf_ci = []\n",
        "\n",
        "for _seed in seeds:\n",
        "  index_train, index_test = train_test_split( range(N), test_size = 0.3, random_state=_seed)\n",
        "  data_train = churn_data.loc[index_train].reset_index(drop=True)\n",
        "  data_test  = churn_data.loc[index_test].reset_index(drop=True)\n",
        "\n",
        "  # Creating the X, T and E inputs\n",
        "  X_trn, X_val = data_train[features], data_test[features]\n",
        "  T_trn, T_val = data_train[time_col], data_test[time_col]\n",
        "  E_trn, E_val = data_train[event_col], data_test[event_col]\n",
        "\n",
        "  rfc = RandomForestClassifier(n_estimators=100, random_state=42)\n",
        "  rfc.fit(X_trn, E_trn)\n",
        "  rfc_ci.append(concordance_index(T_val, rfc.predict_proba(X_val)[:,0], E_val))\n",
        "\n",
        "  rfr1 = RandomForestRegressor(n_estimators=100, random_state=42)\n",
        "  X_trn1 = X_trn[E_trn == True]\n",
        "  T_trn1 = T_trn[E_trn == True]\n",
        "  E_trn1 = E_trn[E_trn == True]\n",
        "  rfr1.fit(X_trn1, T_trn1)\n",
        "  rfr1_ci.append(concordance_index(T_val, rfr1.predict(X_val), E_val))\n",
        "\n",
        "  rfr2 = RandomForestRegressor(n_estimators=100, random_state=42)\n",
        "  rfr2.fit(X_trn, T_trn)\n",
        "  rfr2_ci.append(concordance_index(T_val, rfr2.predict(X_val), E_val))\n",
        "\n",
        "  rsf = RandomSurvivalForestModel(num_trees=100)\n",
        "  rsf.fit(X_trn, T_trn, E_trn, seed=42)\n",
        "  rsf_ci.append(concordance_index(T_val, -rsf.predict_risk(X_val), E_val))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jbr5W5AyZA9r",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 513
        },
        "outputId": "2e69e5e5-a6f7-4fc8-c372-165e343e3dec"
      },
      "source": [
        "data = [rfc_ci, rfr1_ci, rfr2_ci, rsf_ci]\n",
        "green_diamond = dict(markerfacecolor='g', marker='D')\n",
        "\n",
        "fig1, ax1 = plt.subplots(figsize=(7,7))\n",
        "ax1.set_title('Concordance indexes')\n",
        "ax1.boxplot(data, notch=False, flierprops=green_diamond)\n",
        "ax1.set_xticklabels(['Random forest classifier', 'Random forest regressor 1', 'Random forest regressor 2', 'Random survival forest'], rotation=70)\n",
        "ax1.set_xlabel('Models')\n",
        "ax1.set_ylabel('Concordance index')\n",
        "plt.tight_layout()\n",
        "plt.savefig('concordance.png', format='png', dpi=100)\n",
        "# plt.show()"
      ],
      "execution_count": 61,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfAAAAHwCAYAAABZrD3mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCTmBm5mZlZATuJmZWQk5gZuZmZXQP0WrmBZvTZSXAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 504x504 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}