ML_example4.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "authorship_tag": "ABX9TyMvf4fOBlei4nnKRgYvWiir",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/jmbanda/CSC4850-Machine-Learning/blob/main/ML_example4.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# ML Examples Part 4\n",
        "\n",
        "In this notebook we will cover different sampling techniques, how to use GridSearchCV to build a pipeline and finally, plotting ROC and PR curves."
      ],
      "metadata": {
        "id": "g9R7tM4Jotwf"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Introduction to Population and Sample\n",
        "\n",
        "To start with, let’s have a look at some basic terminology. It is important to learn the concepts of population and sample. The **population** is the set of all observations (individuals, objects, events, or procedures) and is usually very large and diverse, whereas a **sample** is a subset of observations from the population that ideally is a true representation of the population.\n",
        "\n",
        "<img src=\"https://miro.medium.com/v2/resize:fit:1400/format:webp/1*VnNrkwNuW2hBKA8DC84Gdg.png\">\n",
        "\n",
        "Given that experimenting with an entire population is either impossible or simply too expensive, researchers or analysts use samples rather than the entire population in their experiments or trials. To make sure that the experimental results are reliable and hold for the entire population, the sample needs to be a true representation of the population. That is, the sample needs to be unbiased."
      ],
      "metadata": {
        "id": "owEbOfxooxh1"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Random Sampling\n",
        "\n",
        "The simplest data sampling technique that creates a random sample from the original population is Random Sampling. In this approach, every sampled observation has the same probability of getting selected during the sample generation process. Random Sampling is usually used when we don’t have any kind of prior information about the target population.\n",
        "\n",
        "For example random selection of 3 individuals from a population of 10 individuals. Here, each individual has an equal chance of getting selected to the sample with a probability of selection of 1/10.\n",
        "\n",
        "<img src=\"https://miro.medium.com/v2/resize:fit:1400/format:webp/1*ZWa1miELJeaGPzRUC9NwqA.png\">"
      ],
      "metadata": {
        "id": "ysGqIVQVpxeE"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Random Sampling: Python Implementation\n",
        "\n",
        "\n",
        "First, we generate random data that will serve as population data. We will, therefore, randomly sample 10K data points from Normal distribution with mean mu = 10 and standard deviation std = 2. After this, we create a Python function called random_sampling() that takes population data and desired sample size and produces as output a random sample."
      ],
      "metadata": {
        "id": "_HX4GgFEp-lz"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "\n",
        "# generating population data following Normal Distribution\n",
        "N = 10000\n",
        "mu = 10\n",
        "std = 2\n",
        "population_df = np.random.normal(mu,std,N)\n",
        "\n",
        "# function that creates random sample \n",
        "def random_sampling(df, n):\n",
        "    random_sample = np.random.choice(df,replace = False, size = n)\n",
        "    return(random_sample)\n",
        "randomSample = random_sampling(population_df, N)\n",
        "randomSample"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pe4GQIO7qD9c",
        "outputId": "9941d923-bb48-4acc-cb43-6a8695268a28"
      },
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([14.78968076,  8.15751783,  8.36155222, ...,  8.34425408,\n",
              "       10.63244894, 11.94724772])"
            ]
          },
          "metadata": {},
          "execution_count": 2
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Systematic Sampling\n",
        "\n",
        "Systematic sampling is defined as a probability sampling approach where the elements from a target population are selected from a random starting point and after a fixed sampling interval.\n",
        "\n",
        "Stated differently, systematic sampling is an extended version of probability sampling techniques in which each member of the group is selected at regular periods to form a sample. We calculate the sampling interval by dividing the entire population size by the desired sample size.\n",
        "\n",
        "Note that, Systematic Sampling usually produces a random sample but is not addressing the bias in the created sample.\n",
        "\n",
        "<img src=\"https://miro.medium.com/v2/resize:fit:1400/format:webp/1*FLyZ66yPDa44EYwTvP7mVQ.png\">"
      ],
      "metadata": {
        "id": "Mr3rnfiKrNcH"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Systematic Sampling: Python Implementation\n",
        "\n",
        "We generate data that serve as population data as in the previous case. We then create a Python function called systematic_sample() that takes population data and interval for the sampling and produces as output a systematic sample."
      ],
      "metadata": {
        "id": "b9FiXpmdrkVo"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "# generating population data following Normal Distribution\n",
        "N = 10000\n",
        "mu = 10\n",
        "std = 2\n",
        "population_df = np.random.normal(mu,std,N)\n",
        "\n",
        "# function that creates random sample using Systematic Sampling\n",
        "def systematic_sampling(df, step):\n",
        "    id = pd.Series(np.arange(1,len(df),1))\n",
        "    df = pd.Series(df)\n",
        "    df_pd = pd.concat([id, df], axis = 1)\n",
        "    df_pd.columns = [\"id\", \"data\"]\n",
        "    # these indices will increase with the step amount not 1\n",
        "    selected_index = np.arange(1,len(df),step)\n",
        "    # using iloc for getting thee data with selected indices\n",
        "    systematic_sampling = df_pd.iloc[selected_index]\n",
        "    return(systematic_sampling)\n",
        "\n",
        "n = 10\n",
        "step = int(N/n)\n",
        "sample = systematic_sampling(population_df, step)\n",
        "sample"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 363
        },
        "id": "x_ULaU3Kr4B1",
        "outputId": "de6c674c-ba97-49cb-dd76-cc598e6f008a"
      },
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "          id       data\n",
              "1        2.0   8.261409\n",
              "1001  1002.0  10.941693\n",
              "2001  2002.0   9.506230\n",
              "3001  3002.0   7.318524\n",
              "4001  4002.0   8.510567\n",
              "5001  5002.0  10.140102\n",
              "6001  6002.0  10.121986\n",
              "7001  7002.0   7.405897\n",
              "8001  8002.0  10.172783\n",
              "9001  9002.0   7.903344"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-aa01550d-70d0-4040-9dcd-db2ec1e588c3\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>id</th>\n",
              "      <th>data</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2.0</td>\n",
              "      <td>8.261409</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1001</th>\n",
              "      <td>1002.0</td>\n",
              "      <td>10.941693</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2001</th>\n",
              "      <td>2002.0</td>\n",
              "      <td>9.506230</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3001</th>\n",
              "      <td>3002.0</td>\n",
              "      <td>7.318524</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4001</th>\n",
              "      <td>4002.0</td>\n",
              "      <td>8.510567</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5001</th>\n",
              "      <td>5002.0</td>\n",
              "      <td>10.140102</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6001</th>\n",
              "      <td>6002.0</td>\n",
              "      <td>10.121986</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7001</th>\n",
              "      <td>7002.0</td>\n",
              "      <td>7.405897</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8001</th>\n",
              "      <td>8002.0</td>\n",
              "      <td>10.172783</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9001</th>\n",
              "      <td>9002.0</td>\n",
              "      <td>7.903344</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-aa01550d-70d0-4040-9dcd-db2ec1e588c3')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-aa01550d-70d0-4040-9dcd-db2ec1e588c3 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-aa01550d-70d0-4040-9dcd-db2ec1e588c3');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Cluster Sampling\n",
        "\n",
        "Cluster sampling is a probability sampling technique where we divide the population into multiple clusters(groups) based on certain clustering criteria. Then we select a random cluster(s) with simple random or systematic sampling techniques. So, in cluster sampling, the entire population is divided into clusters or segments and then cluster(s) are randomly selected.\n",
        "\n",
        "For example, if you want to conduct an experience evaluating the performance of sophomores in business education across Europe. It is impossible to conduct an experiment that involves a student in every university across the EU. Instead, by using Cluster Sampling, we can group the universities from each country into one cluster. These clusters then define all the sophomore student population in the EU. Next, you can use simple random sampling or systematic sampling and randomly select cluster(s) for the purposes of your research study.\n",
        "\n",
        "Note that, Systematic Sampling usually produces a random sample but is not addressing the bias in the created sample.\n",
        "\n",
        "<img src=\"https://miro.medium.com/v2/resize:fit:1400/format:webp/1*4GTPOFJmK3U8lY9H2nw5nw.png\">\n",
        "\n"
      ],
      "metadata": {
        "id": "XNioiksYsJBe"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Cluster Sampling: Python Implementation\n",
        "\n",
        "First, we generate data that will serve as population data with 10K observations, and this data consists of the following 4 variables:\n",
        "\n",
        "Price: generated using Uniform distribution,\n",
        "Id\n",
        "event_type: which is a categorical variable with 3 possible values {type1, type2, type3}\n",
        "click: binary variable taking values {0: no click, 1: click}"
      ],
      "metadata": {
        "id": "vWMwXmOVs-_5"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "\n",
        "# Generating Population data \n",
        "price_vb = pd.Series(np.random.uniform(1,4,size = N))\n",
        "id = pd.Series(np.arange(0,len(price_vb),1))\n",
        "event_type = pd.Series(np.random.choice([\"type1\",\"type2\",\"type3\"],size = len(price_vb)))\n",
        "click = pd.Series(np.random.choice([0,1],size = len(price_vb)))\n",
        "df = pd.concat([id,price_vb,event_type, click],axis = 1)\n",
        "df.columns = [\"id\",\"price\",\"event_type\", \"click\"]\n",
        "df"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 424
        },
        "id": "Cgt6QvMqtKWF",
        "outputId": "a87a21d5-8e70-40c5-bcc6-ffcdeca8e912"
      },
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "        id     price event_type  click\n",
              "0        0  3.348591      type3      0\n",
              "1        1  1.250109      type2      1\n",
              "2        2  2.686037      type3      0\n",
              "3        3  2.691641      type1      1\n",
              "4        4  2.589645      type2      1\n",
              "...    ...       ...        ...    ...\n",
              "9995  9995  1.243976      type2      0\n",
              "9996  9996  2.246366      type3      0\n",
              "9997  9997  2.749073      type3      1\n",
              "9998  9998  2.714811      type2      0\n",
              "9999  9999  3.775007      type3      1\n",
              "\n",
              "[10000 rows x 4 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-8d245e73-2840-4b97-9559-27a57c5d9cb1\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>id</th>\n",
              "      <th>price</th>\n",
              "      <th>event_type</th>\n",
              "      <th>click</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>3.348591</td>\n",
              "      <td>type3</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1</td>\n",
              "      <td>1.250109</td>\n",
              "      <td>type2</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2</td>\n",
              "      <td>2.686037</td>\n",
              "      <td>type3</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>3</td>\n",
              "      <td>2.691641</td>\n",
              "      <td>type1</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>4</td>\n",
              "      <td>2.589645</td>\n",
              "      <td>type2</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9995</th>\n",
              "      <td>9995</td>\n",
              "      <td>1.243976</td>\n",
              "      <td>type2</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9996</th>\n",
              "      <td>9996</td>\n",
              "      <td>2.246366</td>\n",
              "      <td>type3</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9997</th>\n",
              "      <td>9997</td>\n",
              "      <td>2.749073</td>\n",
              "      <td>type3</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9998</th>\n",
              "      <td>9998</td>\n",
              "      <td>2.714811</td>\n",
              "      <td>type2</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9999</th>\n",
              "      <td>9999</td>\n",
              "      <td>3.775007</td>\n",
              "      <td>type3</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>10000 rows × 4 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-8d245e73-2840-4b97-9559-27a57c5d9cb1')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-8d245e73-2840-4b97-9559-27a57c5d9cb1 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-8d245e73-2840-4b97-9559-27a57c5d9cb1');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Then the function `get_clustered_Sample()` takes as inputs the original data, the amount of observations per cluster, and a number of clusters you want to select, and produces as output a clustered sample."
      ],
      "metadata": {
        "id": "_EEQQgeetPqC"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def get_clustered_Sample(df, n_per_cluster, num_select_clusters):\n",
        "    N = len(df)\n",
        "    K = int(N/n_per_cluster)\n",
        "    data = None\n",
        "    for k in range(K):\n",
        "        sample_k = df.sample(n_per_cluster)\n",
        "        sample_k[\"cluster\"] = np.repeat(k,len(sample_k))\n",
        "        df = df.drop(index = sample_k.index)\n",
        "        data = pd.concat([data,sample_k],axis = 0)\n",
        "\n",
        "    random_chosen_clusters = np.random.randint(0,K,size = num_select_clusters)\n",
        "    samples = data[data.cluster.isin(random_chosen_clusters)]\n",
        "    return(samples)\n",
        "\n",
        "sample = get_clustered_Sample(df = df, n_per_cluster = 100, num_select_clusters = 2)\n",
        "sample"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 424
        },
        "id": "lVMiu7UQtViX",
        "outputId": "aa9e6a93-bb96-4165-ca40-c4649079f22f"
      },
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "        id     price event_type  click  cluster\n",
              "9226  9226  3.727015      type3      1       34\n",
              "7140  7140  1.419832      type2      1       34\n",
              "8641  8641  1.543819      type1      0       34\n",
              "4193  4193  1.102810      type3      1       34\n",
              "4782  4782  1.495790      type2      0       34\n",
              "...    ...       ...        ...    ...      ...\n",
              "2485  2485  2.769444      type3      1       96\n",
              "4052  4052  2.270999      type2      1       96\n",
              "4343  4343  1.173676      type2      1       96\n",
              "8690  8690  3.849399      type3      1       96\n",
              "2212  2212  1.880550      type2      1       96\n",
              "\n",
              "[200 rows x 5 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-870c0bc5-a4c1-47c1-b468-b4a40d52054d\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>id</th>\n",
              "      <th>price</th>\n",
              "      <th>event_type</th>\n",
              "      <th>click</th>\n",
              "      <th>cluster</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>9226</th>\n",
              "      <td>9226</td>\n",
              "      <td>3.727015</td>\n",
              "      <td>type3</td>\n",
              "      <td>1</td>\n",
              "      <td>34</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7140</th>\n",
              "      <td>7140</td>\n",
              "      <td>1.419832</td>\n",
              "      <td>type2</td>\n",
              "      <td>1</td>\n",
              "      <td>34</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8641</th>\n",
              "      <td>8641</td>\n",
              "      <td>1.543819</td>\n",
              "      <td>type1</td>\n",
              "      <td>0</td>\n",
              "      <td>34</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4193</th>\n",
              "      <td>4193</td>\n",
              "      <td>1.102810</td>\n",
              "      <td>type3</td>\n",
              "      <td>1</td>\n",
              "      <td>34</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4782</th>\n",
              "      <td>4782</td>\n",
              "      <td>1.495790</td>\n",
              "      <td>type2</td>\n",
              "      <td>0</td>\n",
              "      <td>34</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2485</th>\n",
              "      <td>2485</td>\n",
              "      <td>2.769444</td>\n",
              "      <td>type3</td>\n",
              "      <td>1</td>\n",
              "      <td>96</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4052</th>\n",
              "      <td>4052</td>\n",
              "      <td>2.270999</td>\n",
              "      <td>type2</td>\n",
              "      <td>1</td>\n",
              "      <td>96</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4343</th>\n",
              "      <td>4343</td>\n",
              "      <td>1.173676</td>\n",
              "      <td>type2</td>\n",
              "      <td>1</td>\n",
              "      <td>96</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8690</th>\n",
              "      <td>8690</td>\n",
              "      <td>3.849399</td>\n",
              "      <td>type3</td>\n",
              "      <td>1</td>\n",
              "      <td>96</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2212</th>\n",
              "      <td>2212</td>\n",
              "      <td>1.880550</td>\n",
              "      <td>type2</td>\n",
              "      <td>1</td>\n",
              "      <td>96</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>200 rows × 5 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-870c0bc5-a4c1-47c1-b468-b4a40d52054d')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-870c0bc5-a4c1-47c1-b468-b4a40d52054d button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-870c0bc5-a4c1-47c1-b468-b4a40d52054d');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Weighted Sampling\n",
        "\n",
        "In some experiments, you might need items sampling probabilities to be according to weights associated with each item, that’s when the proportions of the type of observations should be taken into account. For example, you might need a sample of queries in a search engine with weight as a number of times these queries have been performed so that the sample can be analyzed for overall impact on the user experience. In this case, Weighted Sampling is much more preferred compared to Random Sampling or Systematic Sampling.\n",
        "\n",
        "Weighted Sampling is a data sampling method with weights, that intends to compensate for the selection of specific observations with unequal probabilities (oversampling), non-coverage, non-responses, and other types of bias. If a biased data set is not adjusted and a simple random sampling type of approach is used instead, then the population descriptors (e.g., mean, median) will be skewed and they will fail to correctly represent the population’s proportion to the population.\n",
        "\n",
        "Weighted Sampling addresses the bias in the sample, by creating a sample that takes into account the proportions of the type of observations in the population. Hence, Weighted Sampling usually produces a random and unbiased sample.\n",
        "\n",
        "<img src=\"https://miro.medium.com/v2/resize:fit:1400/format:webp/1*zNBVo75a4FGOLo-_ELyiyA.png\">"
      ],
      "metadata": {
        "id": "LF9066FTvBHO"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Weighted Sampling: Python Implementation\n",
        "\n",
        "The function get_weighted_sample() takes as inputs the original data, and the desired sample size, and produces as output a weighted sample. Note that, the proportions, in this case, are defined based on the click event. That is, we compute the proportion of data points that had click events of 1 (let’s say X%) and 0 (Y%, where Y% = 100-X%), then we generate a random sample such that, the sample will also contain X% observations with click = 1 and Y% observations with click = 0."
      ],
      "metadata": {
        "id": "HVq5EaplvQPy"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def get_weighted_sample(df,n):\n",
        "    def get_class_prob(x):\n",
        "        weight_x = int(np.rint(n * len(x[x.click != 0]) / len(df[df.click != 0])))\n",
        "        sampled_x = x.sample(weight_x).reset_index(drop=True)\n",
        "        return (sampled_x)\n",
        "        # we are grouping by the target class we use for the proportions\n",
        "\n",
        "    weighted_sample = df.groupby('event_type').apply(get_class_prob)\n",
        "    print(weighted_sample[\"event_type\"].value_counts())\n",
        "    return (weighted_sample)\n",
        "\n",
        "sample = get_weighted_sample(df,100)\n",
        "sample"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 524
        },
        "id": "jbFAmJQOvVXE",
        "outputId": "ec885665-0906-410f-ef46-315dc4e533ef"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "type3    35\n",
            "type1    33\n",
            "type2    33\n",
            "Name: event_type, dtype: int64\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "                 id     price event_type  click\n",
              "event_type                                     \n",
              "type1      0   9691  2.616468      type1      1\n",
              "           1   7359  1.126266      type1      0\n",
              "           2   5969  1.938191      type1      0\n",
              "           3   8152  3.651034      type1      1\n",
              "           4    957  2.540044      type1      0\n",
              "...             ...       ...        ...    ...\n",
              "type3      30  8891  3.092970      type3      0\n",
              "           31   470  2.109674      type3      1\n",
              "           32  2120  3.417902      type3      1\n",
              "           33  6715  2.536572      type3      0\n",
              "           34  8656  3.670663      type3      0\n",
              "\n",
              "[101 rows x 4 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-c67c7c99-31ee-4ee1-9e7c-e4a60ddad07d\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th>id</th>\n",
              "      <th>price</th>\n",
              "      <th>event_type</th>\n",
              "      <th>click</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>event_type</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th rowspan=\"5\" valign=\"top\">type1</th>\n",
              "      <th>0</th>\n",
              "      <td>9691</td>\n",
              "      <td>2.616468</td>\n",
              "      <td>type1</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>7359</td>\n",
              "      <td>1.126266</td>\n",
              "      <td>type1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>5969</td>\n",
              "      <td>1.938191</td>\n",
              "      <td>type1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>8152</td>\n",
              "      <td>3.651034</td>\n",
              "      <td>type1</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>957</td>\n",
              "      <td>2.540044</td>\n",
              "      <td>type1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th rowspan=\"5\" valign=\"top\">type3</th>\n",
              "      <th>30</th>\n",
              "      <td>8891</td>\n",
              "      <td>3.092970</td>\n",
              "      <td>type3</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>31</th>\n",
              "      <td>470</td>\n",
              "      <td>2.109674</td>\n",
              "      <td>type3</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>32</th>\n",
              "      <td>2120</td>\n",
              "      <td>3.417902</td>\n",
              "      <td>type3</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>33</th>\n",
              "      <td>6715</td>\n",
              "      <td>2.536572</td>\n",
              "      <td>type3</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>34</th>\n",
              "      <td>8656</td>\n",
              "      <td>3.670663</td>\n",
              "      <td>type3</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>101 rows × 4 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-c67c7c99-31ee-4ee1-9e7c-e4a60ddad07d')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-c67c7c99-31ee-4ee1-9e7c-e4a60ddad07d button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-c67c7c99-31ee-4ee1-9e7c-e4a60ddad07d');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Stratified Sampling\n",
        "\n",
        "Stratified Sampling is a data sampling approach, where we divide a population into homogeneous subpopulations called strata based on specific characteristics (e.g., age, race, gender identity, location, event type etc.).\n",
        "\n",
        "Every member of the population studied should be in exactly one stratum. Each stratum is then sampled using Cluster Sampling, allowing data scientists to estimate statistical measures for each sub-population. We rely on Stratified Sampling when the populations’ characteristics are diverse and we want to ensure that every characteristic is properly represented in the sample.\n",
        "\n",
        "So, Stratified Sampling, is simply, the combination of Clustered Sampling and Weighted Sampling.\n",
        "\n",
        "<img src=\"https://miro.medium.com/v2/resize:fit:1400/format:webp/1*iVf_CmE2kERqN9it8UI6Tg.png\">"
      ],
      "metadata": {
        "id": "CAZ1imNduaab"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Stratified Sampling: Python Implementation\n",
        "\n",
        "The function get_stratified_sample() takes as inputs the original data, the desired sample size, the number of clusters needed, and it produces as output a stratified sample. Note that, this function, firstly performs weighted sampling using the click event. Secondly, it performs clustered sampling using the event_type."
      ],
      "metadata": {
        "id": "izcmO0Kcu4Nt"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def get_startified_sample(df,n,num_clusters_needed):\n",
        "    N = len(df)\n",
        "    num_obs_per_cluster = int(N/n)\n",
        "    K = int(N/num_obs_per_cluster)\n",
        "\n",
        "    def get_weighted_sample(df,num_obs_per_cluster):\n",
        "        def get_sample_per_class(x):\n",
        "            n_x = int(np.rint(num_obs_per_cluster*len(x[x.click !=0])/len(df[df.click !=0])))\n",
        "            sample_x = x.sample(n_x)\n",
        "            return(sample_x)\n",
        "        weighted_sample = df.groupby(\"event_type\").apply(get_sample_per_class)\n",
        "        return(weighted_sample)\n",
        "\n",
        "    stratas = None\n",
        "    for k in range(K):\n",
        "        weighted_sample_k = get_weighted_sample(df,num_obs_per_cluster).reset_index(drop = True)\n",
        "        weighted_sample_k[\"cluster\"] = np.repeat(k,len(weighted_sample_k))\n",
        "        stratas = pd.concat([stratas, weighted_sample_k],axis = 0)\n",
        "        df.drop(index = weighted_sample_k.index)\n",
        "    selected_strata_clusters = np.random.randint(0,K,size = num_clusters_needed)\n",
        "    stratified_samples = stratas[stratas.cluster.isin(selected_strata_clusters)]\n",
        "    return(stratified_samples)\n",
        "\n",
        "sample = get_startified_sample(df = df,n = 100,num_clusters_needed = 2)\n",
        "sample"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 424
        },
        "id": "opaGWZ_0vzH6",
        "outputId": "6c2c94d4-9d3e-4cd5-b203-6394a616c219"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "       id     price event_type  click  cluster\n",
              "0    9888  3.124357      type1      1       13\n",
              "1    8516  1.475911      type1      1       13\n",
              "2    3124  3.225580      type1      0       13\n",
              "3    9749  1.070361      type1      1       13\n",
              "4    8255  3.583826      type1      0       13\n",
              "..    ...       ...        ...    ...      ...\n",
              "96    846  3.692454      type3      1       67\n",
              "97   3970  3.764113      type3      0       67\n",
              "98    841  1.075733      type3      1       67\n",
              "99   3545  3.049979      type3      1       67\n",
              "100  2183  2.793530      type3      1       67\n",
              "\n",
              "[202 rows x 5 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-2d93195b-f57c-4d01-9283-7aef95498df3\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>id</th>\n",
              "      <th>price</th>\n",
              "      <th>event_type</th>\n",
              "      <th>click</th>\n",
              "      <th>cluster</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>9888</td>\n",
              "      <td>3.124357</td>\n",
              "      <td>type1</td>\n",
              "      <td>1</td>\n",
              "      <td>13</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>8516</td>\n",
              "      <td>1.475911</td>\n",
              "      <td>type1</td>\n",
              "      <td>1</td>\n",
              "      <td>13</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>3124</td>\n",
              "      <td>3.225580</td>\n",
              "      <td>type1</td>\n",
              "      <td>0</td>\n",
              "      <td>13</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>9749</td>\n",
              "      <td>1.070361</td>\n",
              "      <td>type1</td>\n",
              "      <td>1</td>\n",
              "      <td>13</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>8255</td>\n",
              "      <td>3.583826</td>\n",
              "      <td>type1</td>\n",
              "      <td>0</td>\n",
              "      <td>13</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>96</th>\n",
              "      <td>846</td>\n",
              "      <td>3.692454</td>\n",
              "      <td>type3</td>\n",
              "      <td>1</td>\n",
              "      <td>67</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>97</th>\n",
              "      <td>3970</td>\n",
              "      <td>3.764113</td>\n",
              "      <td>type3</td>\n",
              "      <td>0</td>\n",
              "      <td>67</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>98</th>\n",
              "      <td>841</td>\n",
              "      <td>1.075733</td>\n",
              "      <td>type3</td>\n",
              "      <td>1</td>\n",
              "      <td>67</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>99</th>\n",
              "      <td>3545</td>\n",
              "      <td>3.049979</td>\n",
              "      <td>type3</td>\n",
              "      <td>1</td>\n",
              "      <td>67</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>100</th>\n",
              "      <td>2183</td>\n",
              "      <td>2.793530</td>\n",
              "      <td>type3</td>\n",
              "      <td>1</td>\n",
              "      <td>67</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>202 rows × 5 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-2d93195b-f57c-4d01-9283-7aef95498df3')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-2d93195b-f57c-4d01-9283-7aef95498df3 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-2d93195b-f57c-4d01-9283-7aef95498df3');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Using GridSearchCV\n",
        "\n"
      ],
      "metadata": {
        "id": "s0x4LqtH4Y5D"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "id": "zIHKVxthDZEa"
      },
      "outputs": [],
      "source": [
        "import random\n",
        "import os\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "from sklearn.svm import LinearSVC\n",
        "from sklearn.model_selection import GridSearchCV, train_test_split\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "id": "IEmq50QgDZEi",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "ff1e0e2e-d15b-4e99-f507-c1e93679c61b"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "--2023-03-06 14:53:35--  http://nube.jccm.es/index.php/s/Zs7SYtZQJ3RQ2H2/download\n",
            "Resolving nube.jccm.es (nube.jccm.es)... 89.107.242.157\n",
            "Connecting to nube.jccm.es (nube.jccm.es)|89.107.242.157|:80... connected.\n",
            "HTTP request sent, awaiting response... 302 Moved Temporarily\n",
            "Location: https://nube.jccm.es/index.php/s/Zs7SYtZQJ3RQ2H2/download [following]\n",
            "--2023-03-06 14:53:36--  https://nube.jccm.es/index.php/s/Zs7SYtZQJ3RQ2H2/download\n",
            "Connecting to nube.jccm.es (nube.jccm.es)|89.107.242.157|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 68786819 (66M) [application/x-compressed]\n",
            "Saving to: ‘creditcard.tgz’\n",
            "\n",
            "creditcard.tgz      100%[===================>]  65.60M   793KB/s    in 86s     \n",
            "\n",
            "2023-03-06 14:55:02 (786 KB/s) - ‘creditcard.tgz’ saved [68786819/68786819]\n",
            "\n"
          ]
        }
      ],
      "source": [
        "if not os.path.isfile('data/creditcard.csv'):\n",
        "    !wget --no-check-certificate --content-disposition http://nube.jccm.es/index.php/s/Zs7SYtZQJ3RQ2H2/download\n",
        "    !tar xzf creditcard.tgz"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "id": "z9Q-YUfBDZEq",
        "tags": [],
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "24241eeb-fd8a-4854-836c-11c45d12339f"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Fraud: 0.173% 492\n",
            "Valid: 99.827% 284315\n",
            "X.shape (1492, 28)  y.shape (1492,)\n",
            "Fraud: 32.976% 492\n",
            "Valid: 67.024% 1000\n"
          ]
        }
      ],
      "source": [
        "random_state=1\n",
        "\n",
        "def load_creditcard(n_examples=0):\n",
        "    df = pd.read_csv('data/creditcard.csv')\n",
        "    print(\"Fraud: {0:.3f}% {1}\".format(df.Class[df.Class == 1].count()*100/df.shape[0], df.Class[df.Class == 1].count()))\n",
        "    print(\"Valid: {0:.3f}% {1}\".format(df.Class[df.Class == 0].count()*100/df.shape[0], df.Class[df.Class == 0].count()))\n",
        "    y = df.Class\n",
        "    X = df.drop(['Class', 'Time', 'Amount'], axis=1).values\n",
        "    if n_examples > 0:\n",
        "        # Take first n_examples samples\n",
        "        X = X[:n_examples, :]\n",
        "        y = y[:n_examples, :]\n",
        "    else:\n",
        "        # Take all the positive samples with a number of random negatives\n",
        "        if n_examples < 0:\n",
        "            Xt = X[(y == 1).ravel()]\n",
        "            yt = y[(y == 1).ravel()]\n",
        "            indices = random.sample(range(X.shape[0]), -1 * n_examples)\n",
        "            X = np.append(Xt, X[indices], axis=0)\n",
        "            y = np.append(yt, y[indices], axis=0)\n",
        "    print(\"X.shape\", X.shape, \" y.shape\", y.shape)\n",
        "    print(\"Fraud: {0:.3f}% {1}\".format(len(y[y == 1])*100/X.shape[0], len(y[y == 1])))\n",
        "    print(\"Valid: {0:.3f}% {1}\".format(len(y[y == 0]) * 100 / X.shape[0], len(y[y == 0])))\n",
        "    Xtrain, Xtest, ytrain, ytest = train_test_split(X, y, train_size=0.7, shuffle=True, random_state=random_state, stratify=y)\n",
        "    return Xtrain, Xtest, ytrain, ytest\n",
        "\n",
        "data = load_creditcard(-1000) \n",
        "#data = load_creditcard(0) # Take all the samples\n",
        "\n",
        "Xtrain = data[0]\n",
        "Xtest = data[1]\n",
        "ytrain = data[2]\n",
        "ytest = data[3]"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.ensemble import RandomForestClassifier\n",
        "\n",
        "# Create a Random Forest Classifier with specified criterion\n",
        "rf_class = RandomForestClassifier(criterion='entropy')\n",
        "\n",
        "# Create the parametergrid\n",
        "param_grid = {\n",
        "    'max_depth':[2, 4, 8, 15],\n",
        "    'max_features':['auto', 'sqrt']\n",
        "}\n",
        "\n",
        "# Create a GridSearchCV object\n",
        "grid_rf_class = GridSearchCV(\n",
        "    estimator=rf_class,\n",
        "    param_grid=param_grid,\n",
        "    scoring='roc_auc',\n",
        "    n_jobs=4,\n",
        "    cv=5,\n",
        "    refit=True,\n",
        "    return_train_score=True\n",
        ")\n",
        "\n",
        "print(grid_rf_class)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Dz1X_aUHawYI",
        "outputId": "86bc124d-663d-4923-8ea1-b66002b44a40"
      },
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "GridSearchCV(cv=5, estimator=RandomForestClassifier(criterion='entropy'),\n",
            "             n_jobs=4,\n",
            "             param_grid={'max_depth': [2, 4, 8, 15],\n",
            "                         'max_features': ['auto', 'sqrt']},\n",
            "             return_train_score=True, scoring='roc_auc')\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oJfbencxfG7T"
      },
      "source": [
        "## Understanding a grid search output\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Y74qc1bAfG7U"
      },
      "source": [
        "### Exploring the grid search results\n",
        "You will now explore the `cv_results_` property of the GridSearchCV object. This is a dictionary that we can read into a pandas DataFrame and contains a lot of useful information about the grid search we just undertook.\n",
        "\n",
        "A reminder of the different column types in this property:\n",
        "\n",
        "- `time_` columns\n",
        "- `param_` columns (one for each hyperparameter) and the singular params column (with all hyperparameter settings)\n",
        "- a `train_score` column for each cv fold including the mean_train_score and std_train_score columns\n",
        "- a `test_score` column for each cv fold including the mean_test_score and std_test_score columns\n",
        "- a `rank_test_score` column with a number from 1 to n (number of iterations) ranking the rows based on their `mean_test_score`"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "id": "tkzsW_H9fG7V",
        "outputId": "8814b5eb-ff30-4994-96d3-3994bf89e1b5",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "   mean_fit_time  std_fit_time  mean_score_time  std_score_time  \\\n",
            "0       0.550368      0.068459         0.032146        0.004491   \n",
            "1       0.562965      0.029085         0.034830        0.002976   \n",
            "2       0.764004      0.062777         0.027176        0.011484   \n",
            "3       0.720883      0.054319         0.031080        0.004391   \n",
            "4       0.922846      0.036812         0.028044        0.000950   \n",
            "5       0.877358      0.071911         0.030115        0.003781   \n",
            "6       0.943463      0.017882         0.030261        0.002776   \n",
            "7       0.763517      0.115416         0.020940        0.006677   \n",
            "\n",
            "  param_max_depth param_max_features  \\\n",
            "0               2               auto   \n",
            "1               2               sqrt   \n",
            "2               4               auto   \n",
            "3               4               sqrt   \n",
            "4               8               auto   \n",
            "5               8               sqrt   \n",
            "6              15               auto   \n",
            "7              15               sqrt   \n",
            "\n",
            "                                      params  split0_test_score  \\\n",
            "0   {'max_depth': 2, 'max_features': 'auto'}           0.986749   \n",
            "1   {'max_depth': 2, 'max_features': 'sqrt'}           0.988302   \n",
            "2   {'max_depth': 4, 'max_features': 'auto'}           0.992650   \n",
            "3   {'max_depth': 4, 'max_features': 'sqrt'}           0.990166   \n",
            "4   {'max_depth': 8, 'max_features': 'auto'}           0.993892   \n",
            "5   {'max_depth': 8, 'max_features': 'sqrt'}           0.993892   \n",
            "6  {'max_depth': 15, 'max_features': 'auto'}           0.993892   \n",
            "7  {'max_depth': 15, 'max_features': 'sqrt'}           0.992547   \n",
            "\n",
            "   split1_test_score  split2_test_score  ...  mean_test_score  std_test_score  \\\n",
            "0           0.953830           0.972774  ...         0.969079        0.011316   \n",
            "1           0.957816           0.967133  ...         0.969048        0.010754   \n",
            "2           0.961387           0.978468  ...         0.974911        0.013095   \n",
            "3           0.965631           0.978157  ...         0.971452        0.013046   \n",
            "4           0.965010           0.978882  ...         0.975492        0.015200   \n",
            "5           0.969048           0.983747  ...         0.976902        0.015949   \n",
            "6           0.969462           0.986749  ...         0.976480        0.015274   \n",
            "7           0.971739           0.982609  ...         0.976258        0.014564   \n",
            "\n",
            "   rank_test_score  split0_train_score  split1_train_score  \\\n",
            "0                7            0.976990            0.980552   \n",
            "1                8            0.976977            0.981188   \n",
            "2                5            0.992357            0.994883   \n",
            "3                6            0.992669            0.995753   \n",
            "4                4            0.999922            0.999994   \n",
            "5                1            0.999851            0.999974   \n",
            "6                2            1.000000            1.000000   \n",
            "7                3            1.000000            1.000000   \n",
            "\n",
            "   split2_train_score  split3_train_score  split4_train_score  \\\n",
            "0            0.977503            0.981734            0.976097   \n",
            "1            0.978464            0.984016            0.977552   \n",
            "2            0.992110            0.994825            0.993815   \n",
            "3            0.992273            0.995974            0.991285   \n",
            "4            0.999961            0.999981            0.999968   \n",
            "5            0.999916            0.999994            0.999935   \n",
            "6            1.000000            1.000000            1.000000   \n",
            "7            1.000000            1.000000            1.000000   \n",
            "\n",
            "   mean_train_score  std_train_score  \n",
            "0          0.978575     2.176619e-03  \n",
            "1          0.979640     2.622181e-03  \n",
            "2          0.993598     1.179485e-03  \n",
            "3          0.993591     1.911036e-03  \n",
            "4          0.999965     2.415982e-05  \n",
            "5          0.999934     4.986197e-05  \n",
            "6          1.000000     4.965068e-17  \n",
            "7          1.000000     0.000000e+00  \n",
            "\n",
            "[8 rows x 22 columns]\n",
            "                                      params\n",
            "0   {'max_depth': 2, 'max_features': 'auto'}\n",
            "1   {'max_depth': 2, 'max_features': 'sqrt'}\n",
            "2   {'max_depth': 4, 'max_features': 'auto'}\n",
            "3   {'max_depth': 4, 'max_features': 'sqrt'}\n",
            "4   {'max_depth': 8, 'max_features': 'auto'}\n",
            "5   {'max_depth': 8, 'max_features': 'sqrt'}\n",
            "6  {'max_depth': 15, 'max_features': 'auto'}\n",
            "7  {'max_depth': 15, 'max_features': 'sqrt'}\n",
            "   mean_fit_time  std_fit_time  mean_score_time  std_score_time  \\\n",
            "5       0.877358      0.071911         0.030115        0.003781   \n",
            "\n",
            "  param_max_depth param_max_features  \\\n",
            "5               8               sqrt   \n",
            "\n",
            "                                     params  split0_test_score  \\\n",
            "5  {'max_depth': 8, 'max_features': 'sqrt'}           0.993892   \n",
            "\n",
            "   split1_test_score  split2_test_score  ...  mean_test_score  std_test_score  \\\n",
            "5           0.969048           0.983747  ...         0.976902        0.015949   \n",
            "\n",
            "   rank_test_score  split0_train_score  split1_train_score  \\\n",
            "5                1            0.999851            0.999974   \n",
            "\n",
            "   split2_train_score  split3_train_score  split4_train_score  \\\n",
            "5            0.999916            0.999994            0.999935   \n",
            "\n",
            "   mean_train_score  std_train_score  \n",
            "5          0.999934          0.00005  \n",
            "\n",
            "[1 rows x 22 columns]\n"
          ]
        }
      ],
      "source": [
        "grid_rf_class.fit(Xtrain, ytrain)\n",
        "\n",
        "# Read the cv_results property into adataframe & print it out\n",
        "cv_results_df = pd.DataFrame(grid_rf_class.cv_results_)\n",
        "print(cv_results_df)\n",
        "\n",
        "# Extract and print the column with a dictionary of hyperparameters used\n",
        "column = cv_results_df.loc[:, [\"params\"]]\n",
        "print(column)\n",
        "\n",
        "# Extract and print the row that had the best mean test score\n",
        "best_row = cv_results_df[cv_results_df['rank_test_score'] == 1]\n",
        "print(best_row)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oNHzVUtHfG7W"
      },
      "source": [
        "### Analyzing the best results\n",
        "At the end of the day, we primarily care about the best performing 'square' in a grid search. Luckily Scikit Learn's `gridSearchCV` objects have a number of parameters that provide key information on just the best square (or row in `cv_results_`).\n",
        "\n",
        "Three properties you will explore are:\n",
        "\n",
        "- `best_score_` – The score (here ROC_AUC) from the best-performing square.\n",
        "- `best_index_` – The index of the row in `cv_results_` containing information on the best-performing square.\n",
        "- `best_params_` – A dictionary of the parameters that gave the best score, for example 'max_depth': 10"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "id": "T3ZQdaHrfG7X",
        "outputId": "a91d8d09-b7de-40a2-adfb-d2bd22cf0247",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0.9769017172086226\n",
            "   mean_fit_time  std_fit_time  mean_score_time  std_score_time  \\\n",
            "5       0.877358      0.071911         0.030115        0.003781   \n",
            "\n",
            "  param_max_depth param_max_features  \\\n",
            "5               8               sqrt   \n",
            "\n",
            "                                     params  split0_test_score  \\\n",
            "5  {'max_depth': 8, 'max_features': 'sqrt'}           0.993892   \n",
            "\n",
            "   split1_test_score  split2_test_score  ...  mean_test_score  std_test_score  \\\n",
            "5           0.969048           0.983747  ...         0.976902        0.015949   \n",
            "\n",
            "   rank_test_score  split0_train_score  split1_train_score  \\\n",
            "5                1            0.999851            0.999974   \n",
            "\n",
            "   split2_train_score  split3_train_score  split4_train_score  \\\n",
            "5            0.999916            0.999994            0.999935   \n",
            "\n",
            "   mean_train_score  std_train_score  \n",
            "5          0.999934          0.00005  \n",
            "\n",
            "[1 rows x 22 columns]\n",
            "8\n"
          ]
        }
      ],
      "source": [
        "# Print out the ROC_AUC score from the best-performing square\n",
        "best_score = grid_rf_class.best_score_\n",
        "print(best_score)\n",
        "\n",
        "# Create a variable from the row related to the best-performing square\n",
        "cv_results_df = pd.DataFrame(grid_rf_class.cv_results_)\n",
        "best_row = cv_results_df.loc[[grid_rf_class.best_index_]]\n",
        "print(best_row)\n",
        "\n",
        "# Get the max_depth parameter from the best-performing square and print\n",
        "best_max_depth = grid_rf_class.best_params_['max_depth']\n",
        "print(best_max_depth)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9XzmrNO2fG7Z"
      },
      "source": [
        "### Using the best results\n",
        "While it is interesting to analyze the results of our grid search, our final goal is practical in nature; we want to make predictions on our test set using our estimator object.\n",
        "\n",
        "We can access this object through the `best_estimator_` property of our grid search object.\n",
        "\n",
        "In this exercise we will take a look inside the `best_estimator_` property and then use this to make predictions on our test set for credit card defaults and generate a variety of scores. Remember to use `predict_proba` rather than `predict` since we need probability values rather than class labels for our roc_auc score. We use a slice `[:,1]` to get probabilities of the positive class."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {
        "id": "qkuaO070fG7Z",
        "outputId": "7b1b09dd-eed0-423c-e40b-7d1e2543a829",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "<class 'sklearn.ensemble._forest.RandomForestClassifier'>\n",
            "[0 0 0 0 0]\n",
            "Confusion Matrix \n",
            " [[296   4]\n",
            " [ 19 129]]\n",
            "ROC-AUC Score \n",
            " 0.9792567567567567\n"
          ]
        }
      ],
      "source": [
        "from sklearn.metrics import confusion_matrix, roc_auc_score\n",
        "\n",
        "# See what type of object the best_estimator_property is\n",
        "print(type(grid_rf_class.best_estimator_))\n",
        "\n",
        "# Create an array of predictions directly using the best_estimator_property\n",
        "predictions = grid_rf_class.best_estimator_.predict(Xtest)\n",
        "\n",
        "# Take a look to confirm it worked, this should be an array of 1's and 0's\n",
        "print(predictions[0:5])\n",
        "\n",
        "# Now create a confusion matrix\n",
        "print(\"Confusion Matrix \\n\", confusion_matrix(ytest, predictions))\n",
        "\n",
        "# Get the ROC-AUC score\n",
        "predictions_proba = grid_rf_class.best_estimator_.predict_proba(Xtest)[:, 1]\n",
        "print(\"ROC-AUC Score \\n\", roc_auc_score(ytest, predictions_proba))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QwY_f1AifG7b"
      },
      "source": [
        "The `.best_estimator_` property is a really powerful property to understand for streamlining your machine learning model building process. You now can run a grid search and seamlessly use the best model from that search to make predictions."
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(\"Best estimator: \", grid_rf_class.best_estimator_)\n",
        "print(\"Best hyperparameters: \", grid_rf_class.best_params_)\n",
        "print(\"Best accuracy: \", grid_rf_class.best_score_)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Kd5spgTucmOL",
        "outputId": "f0c61b43-36dc-4c20-b526-021c972b2a35"
      },
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Best estimator:  RandomForestClassifier(criterion='entropy', max_depth=8)\n",
            "Best hyperparameters:  {'max_depth': 8, 'max_features': 'sqrt'}\n",
            "Best accuracy:  0.9769017172086226\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Plotting ROC curves"
      ],
      "metadata": {
        "id": "5gmlZ0T4dhL3"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "from itertools import cycle\n",
        "\n",
        "from sklearn import svm, datasets\n",
        "from sklearn.metrics import roc_curve, auc\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import label_binarize\n",
        "from sklearn.multiclass import OneVsRestClassifier\n",
        "from scipy import interp\n",
        "\n",
        "# Import some data to play with\n",
        "iris = datasets.load_iris()\n",
        "X = iris.data\n",
        "y = iris.target\n",
        "\n",
        "# Binarize the output\n",
        "y = label_binarize(y, classes=[0, 1, 2])\n",
        "n_classes = y.shape[1]\n",
        "\n",
        "# Add noisy features to make the problem harder\n",
        "random_state = np.random.RandomState(0)\n",
        "n_samples, n_features = X.shape\n",
        "X = np.c_[X, random_state.randn(n_samples, 200 * n_features)]\n",
        "\n",
        "# shuffle and split training and test sets\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=.5,\n",
        "                                                    random_state=0)\n",
        "\n",
        "# Learn to predict each class against the other\n",
        "classifier = OneVsRestClassifier(svm.SVC(kernel='linear', probability=True,\n",
        "                                 random_state=random_state))\n",
        "y_score = classifier.fit(X_train, y_train).decision_function(X_test)\n",
        "\n",
        "# Compute ROC curve and ROC area for each class\n",
        "fpr = dict()\n",
        "tpr = dict()\n",
        "roc_auc = dict()\n",
        "for i in range(n_classes):\n",
        "    fpr[i], tpr[i], _ = roc_curve(y_test[:, i], y_score[:, i])\n",
        "    roc_auc[i] = auc(fpr[i], tpr[i])\n",
        "\n",
        "# Compute micro-average ROC curve and ROC area\n",
        "fpr[\"micro\"], tpr[\"micro\"], _ = roc_curve(y_test.ravel(), y_score.ravel())\n",
        "roc_auc[\"micro\"] = auc(fpr[\"micro\"], tpr[\"micro\"])"
      ],
      "metadata": {
        "id": "5SXWKf-rdIsH"
      },
      "execution_count": 21,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nKlzEhRydAYf"
      },
      "source": [
        "Plot of a ROC curve for a specific class\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 22,
      "metadata": {
        "id": "SjPuFhrYdAYg",
        "outputId": "316b240e-f20f-4a5f-8f8f-72cb0096a20d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "plt.figure()\n",
        "lw = 2\n",
        "plt.plot(fpr[2], tpr[2], color='darkorange',\n",
        "         lw=lw, label='ROC curve (area = %0.2f)' % roc_auc[2])\n",
        "plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--')\n",
        "plt.xlim([0.0, 1.0])\n",
        "plt.ylim([0.0, 1.05])\n",
        "plt.xlabel('False Positive Rate')\n",
        "plt.ylabel('True Positive Rate')\n",
        "plt.title('Receiver operating characteristic example')\n",
        "plt.legend(loc=\"lower right\")\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WbPZphe3dAYh"
      },
      "source": [
        "Plot ROC curves for the multiclass problem\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 24,
      "metadata": {
        "id": "TLs1t3kzdAYh",
        "outputId": "0a4e447d-ead0-4792-ab67-001c123a6efd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "# Compute macro-average ROC curve and ROC area\n",
        "\n",
        "# First aggregate all false positive rates\n",
        "all_fpr = np.unique(np.concatenate([fpr[i] for i in range(n_classes)]))\n",
        "\n",
        "# Then interpolate all ROC curves at this points\n",
        "mean_tpr = np.zeros_like(all_fpr)\n",
        "for i in range(n_classes):\n",
        "    mean_tpr += np.interp(all_fpr, fpr[i], tpr[i])\n",
        "\n",
        "# Finally average it and compute AUC\n",
        "mean_tpr /= n_classes\n",
        "\n",
        "fpr[\"macro\"] = all_fpr\n",
        "tpr[\"macro\"] = mean_tpr\n",
        "roc_auc[\"macro\"] = auc(fpr[\"macro\"], tpr[\"macro\"])\n",
        "\n",
        "# Plot all ROC curves\n",
        "plt.figure()\n",
        "plt.plot(fpr[\"micro\"], tpr[\"micro\"],\n",
        "         label='micro-average ROC curve (area = {0:0.2f})'\n",
        "               ''.format(roc_auc[\"micro\"]),\n",
        "         color='deeppink', linestyle=':', linewidth=4)\n",
        "\n",
        "plt.plot(fpr[\"macro\"], tpr[\"macro\"],\n",
        "         label='macro-average ROC curve (area = {0:0.2f})'\n",
        "               ''.format(roc_auc[\"macro\"]),\n",
        "         color='navy', linestyle=':', linewidth=4)\n",
        "\n",
        "colors = cycle(['aqua', 'darkorange', 'cornflowerblue'])\n",
        "for i, color in zip(range(n_classes), colors):\n",
        "    plt.plot(fpr[i], tpr[i], color=color, lw=lw,\n",
        "             label='ROC curve of class {0} (area = {1:0.2f})'\n",
        "             ''.format(i, roc_auc[i]))\n",
        "\n",
        "plt.plot([0, 1], [0, 1], 'k--', lw=lw)\n",
        "plt.xlim([0.0, 1.0])\n",
        "plt.ylim([0.0, 1.05])\n",
        "plt.xlabel('False Positive Rate')\n",
        "plt.ylabel('True Positive Rate')\n",
        "plt.title('Some extension of Receiver operating characteristic to multi-class')\n",
        "plt.legend(loc=\"lower right\")\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Plotting Precision Recall Curves"
      ],
      "metadata": {
        "id": "XyKCRp1dfa2Q"
      }
    },
    {
      "metadata": {
        "id": "I4b1HAHx5lP-"
      },
      "cell_type": "code",
      "source": [
        "# Setup\n",
        "import matplotlib.pyplot as plt\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "\n",
        "from scipy.stats import norm\n",
        "from sklearn.metrics import precision_recall_curve\n",
        "from sklearn.metrics import classification_report\n",
        "from sklearn.metrics import roc_curve, auc"
      ],
      "execution_count": 25,
      "outputs": []
    },
    {
      "metadata": {
        "id": "-PGsieux5lQD"
      },
      "cell_type": "markdown",
      "source": [
        "## Generate normally distributed numpy arrays for two equally balanced classes"
      ]
    },
    {
      "metadata": {
        "id": "IIK8xO395lQH",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "53edf99a-9448-4379-9588-2cc2fb4a2f4f"
      },
      "cell_type": "code",
      "source": [
        "def normal_distribution(mu, sigma, N, bins, label, plt_color, is_plot=True):\n",
        "  # Simulate score data from a normal distribution based on: mu, sigma and N\n",
        "  scores = np.sort(norm.rvs(loc=mu, scale=sigma, size=N))\n",
        "  \n",
        "  # Get bell curve for reference\n",
        "  x = np.linspace(-4, 4, N)\n",
        "  bell = norm.pdf(x) * N * max(norm.pdf(x))\n",
        "  \n",
        "  if is_plot:\n",
        "    plt.hist(scores, density=0, color=plt_color, alpha=.5, bins=bins)\n",
        "    plt.plot(x+mu, bell, '--', label=label, color=plt_color)\n",
        "  \n",
        "  return scores, x, bell\n",
        "\n",
        "BINS = 20\n",
        "\n",
        "plt.figure()\n",
        "\n",
        "# Distribution for Negative class labels\n",
        "N1 = 10000\n",
        "mu1 = -1  # distribution is offset by -1 from the centre\n",
        "sigma1 = 1\n",
        "label1 = 'Negative class labels'\n",
        "scores1, x1, bell1 = normal_distribution(mu1, sigma1, N1, BINS, label1, plt_color='red')\n",
        "\n",
        "# Distribution for Positive class labels\n",
        "N2 = 10000\n",
        "mu2 = 0\n",
        "sigma2 = 1\n",
        "label2 = 'Positive class labels'\n",
        "scores2, x2, bell2 = normal_distribution(mu2, sigma2, N2, BINS, label2, plt_color='green')\n",
        "\n",
        "# Plot\n",
        "plt.title(\"Normal distribution plots of Negative and Positive class labels\")\n",
        "plt.xlabel('Score')\n",
        "plt.ylabel('Count')\n",
        "plt.grid(True)\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "ex6rutLX5lQT"
      },
      "cell_type": "markdown",
      "source": [
        "# Build two lists for (i) all the true labels (ii) all the score values"
      ]
    },
    {
      "metadata": {
        "id": "FfIVOfol5lQU",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "e02545e1-906f-4773-8b1d-fccb3ff12c24"
      },
      "cell_type": "code",
      "source": [
        "# Append the true class labels to the list of negative class labels\n",
        "y_true = N1*[0] + N2*[1] \n",
        "y_score = list(scores1) + list(scores2)\n",
        "y_score = [np.round(y,2) for y in y_score]\n",
        "\n",
        "print(\"N1, N2:\", N1, N2)\n",
        "print(\"len(y_true):\", len(y_true))\n",
        "assert(len(y_true) == len(y_score))\n",
        "print(\"---\")\n",
        "\n",
        "print(\"Mid 10 values (right most negative red scores, left most positive green scores:\")\n",
        "print(\"y_true[N1-5:N1+5]:\", y_true[N1-5:N1+5])\n",
        "print(\"y_score[N1-5:N1+5]:\", y_score[N1-5:N1+5])\n"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "N1, N2: 10000 10000\n",
            "len(y_true): 20000\n",
            "---\n",
            "Mid 10 values (right most negative red scores, left most positive green scores:\n",
            "y_true[N1-5:N1+5]: [0, 0, 0, 0, 0, 1, 1, 1, 1, 1]\n",
            "y_score[N1-5:N1+5]: [2.26, 2.37, 2.39, 2.55, 2.83, -3.69, -3.49, -3.4, -3.37, -3.29]\n"
          ]
        }
      ]
    },
    {
      "metadata": {
        "id": "GofF3F6o9TeW"
      },
      "cell_type": "markdown",
      "source": [
        "  ## Print confusion matrix & classification report for different classification thresholds"
      ]
    },
    {
      "metadata": {
        "id": "TMX26jW_9TqW",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "c4243ee9-7d58-4caa-d168-e8b67db8773c"
      },
      "cell_type": "code",
      "source": [
        "for threshold in [-1, -0.5, 0]:  # test 3 score thresholds which are used to determine if a class is predicted to be 0 or 1\n",
        "  print(\"threshold:\", threshold)\n",
        "  print(\"----------\")\n",
        "  y_pred = [0 if y < threshold else 1 for y in y_score]\n",
        "\n",
        "  # from sklearn.metrics import confusion_matrix\n",
        "  # cm = confusion_matrix(y_true, y_pred)\n",
        "\n",
        "  # Pandas 'crosstab' displays a better formated confusion matrix than the one in sklearn\n",
        "  cm = pd.crosstab(pd.Series(y_true), pd.Series(y_pred), rownames=['Reality'], colnames=['Predicted'], margins=True)\n",
        "  print(cm) \n",
        "\n",
        "  print()\n",
        "  print(\"Classification report:\")\n",
        "  print(classification_report(y_true, y_pred))\n",
        "  print()"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "threshold: -1\n",
            "----------\n",
            "Predicted     0      1    All\n",
            "Reality                      \n",
            "0          4930   5070  10000\n",
            "1          1601   8399  10000\n",
            "All        6531  13469  20000\n",
            "\n",
            "Classification report:\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.75      0.49      0.60     10000\n",
            "           1       0.62      0.84      0.72     10000\n",
            "\n",
            "    accuracy                           0.67     20000\n",
            "   macro avg       0.69      0.67      0.66     20000\n",
            "weighted avg       0.69      0.67      0.66     20000\n",
            "\n",
            "\n",
            "threshold: -0.5\n",
            "----------\n",
            "Predicted     0      1    All\n",
            "Reality                      \n",
            "0          6809   3191  10000\n",
            "1          3150   6850  10000\n",
            "All        9959  10041  20000\n",
            "\n",
            "Classification report:\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.68      0.68      0.68     10000\n",
            "           1       0.68      0.69      0.68     10000\n",
            "\n",
            "    accuracy                           0.68     20000\n",
            "   macro avg       0.68      0.68      0.68     20000\n",
            "weighted avg       0.68      0.68      0.68     20000\n",
            "\n",
            "\n",
            "threshold: 0\n",
            "----------\n",
            "Predicted      0     1    All\n",
            "Reality                      \n",
            "0           8363  1637  10000\n",
            "1           5011  4989  10000\n",
            "All        13374  6626  20000\n",
            "\n",
            "Classification report:\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.63      0.84      0.72     10000\n",
            "           1       0.75      0.50      0.60     10000\n",
            "\n",
            "    accuracy                           0.67     20000\n",
            "   macro avg       0.69      0.67      0.66     20000\n",
            "weighted avg       0.69      0.67      0.66     20000\n",
            "\n",
            "\n"
          ]
        }
      ]
    },
    {
      "metadata": {
        "id": "Afe1Op8EiYuA"
      },
      "cell_type": "markdown",
      "source": [
        "## Precision-Recall curves"
      ]
    },
    {
      "metadata": {
        "id": "ladaDy_LiZJy",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 251
        },
        "outputId": "7395c4e5-5e13-4ada-8cb0-e087df713c6a"
      },
      "cell_type": "code",
      "source": [
        "# Get ROC curve FPR and TPR from true labels vs score values\n",
        "fpr, tpr, _ = roc_curve(y_true, y_score)\n",
        "\n",
        "# Calculate ROC Area Under the Curve (AUC) from FPR and TPR data points\n",
        "roc_auc = auc(fpr, tpr)\n",
        "\n",
        "# Calculate precision and recall from true labels vs score values\n",
        "precision, recall, _ = precision_recall_curve(y_true, y_score)\n",
        "\n",
        "plt.figure(figsize=(8, 3))\n",
        "\n",
        "plt.subplot(1,2,1)\n",
        "lw = 2\n",
        "plt.plot(fpr, tpr, color='darkorange', lw=lw, label='ROC curve (area = %0.4f)' % roc_auc)\n",
        "plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--')\n",
        "plt.xlim([0.0, 1.0])\n",
        "plt.ylim([0.0, 1.05])\n",
        "plt.xlabel('False Positive Rate')\n",
        "plt.ylabel('True Positive Rate')\n",
        "plt.title('ROC Curve')\n",
        "plt.legend(loc=\"lower right\")\n",
        "plt.grid(True)\n",
        "\n",
        "plt.subplot(1,2,2)\n",
        "plt.step(recall, precision, color='orange', where='post')\n",
        "# plt.fill_between(recall, precision, step='post', alpha=0.5, color='orange')\n",
        "plt.xlabel('Recall')\n",
        "plt.ylabel('Precision')\n",
        "plt.ylim([0.0, 1.05])\n",
        "plt.xlim([0.0, 1.0])\n",
        "plt.title('Precision Recall Curve')\n",
        "plt.grid(True)\n",
        "\n",
        "left  = 0.125  # the left side of the subplots of the figure\n",
        "right = 0.9    # the right side of the subplots of the figure\n",
        "bottom = 0.1   # the bottom of the subplots of the figure\n",
        "top = 0.9      # the top of the subplots of the figure\n",
        "wspace = 0.5   # the amount of width reserved for blank space between subplots\n",
        "hspace = 0.2   # the amount of height reserved for white space between subplots\n",
        "plt.subplots_adjust(left, bottom, right, top, wspace, hspace)\n",
        "plt.show()"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x216 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Sources\n",
        "\n",
        "https://towardsdatascience.com/data-sampling-methods-in-python-a4400628ea1b\n",
        "\n",
        "https://colab.research.google.com/github/rpi-techfundamentals/spring2019-materials/blob/master/05-intro-modeling/04-svm-roc.ipynb\n",
        "\n"
      ],
      "metadata": {
        "id": "-sxde4fCWR6E"
      }
    }
  ]
}