Shapley_demo.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Shapley_demo.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CToXyudcDrcg"
      },
      "source": [
        "# Intro\n",
        "This notebook shows how to utilise the [shap python library](https://pypi.org/project/shap/) to interpret tree based models. The aim is to provide a practical reference for use of shap on common tree based models.\n",
        "\n",
        "Each of the implemented models is trained to predict house price on the California Housing dataset. Each is tuned to provide a good representation of an effective model.\n",
        "\n",
        "The built in feature importance measures are then plotted for the trained models. These can then be compared to the plots generated by the shap library.\n",
        "\n",
        "Main references:\n",
        "1.   https://github.com/slundberg/shap\n",
        "2.   https://towardsdatascience.com/interpretable-machine-learning-with-xgboost-9ec80d148d27\n",
        "3.   https://towardsdatascience.com/explain-your-model-with-the-shap-values-bc36aac4de3d\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DDjcBs_FDvTJ"
      },
      "source": [
        "# Dependencies"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Sbc5Nnm7vRnG",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "1efa8333-77fd-410b-db40-15fac10c05b5"
      },
      "source": [
        "!pip install shap"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting shap\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/b9/f4/c5b95cddae15be80f8e58b25edceca105aa83c0b8c86a1edad24a6af80d3/shap-0.39.0.tar.gz (356kB)\n",
            "\r\u001b[K     |█                               | 10kB 23.5MB/s eta 0:00:01\r\u001b[K     |█▉                              | 20kB 31.0MB/s eta 0:00:01\r\u001b[K     |██▊                             | 30kB 22.1MB/s eta 0:00:01\r\u001b[K     |███▊                            | 40kB 25.7MB/s eta 0:00:01\r\u001b[K     |████▋                           | 51kB 24.0MB/s eta 0:00:01\r\u001b[K     |█████▌                          | 61kB 17.1MB/s eta 0:00:01\r\u001b[K     |██████▍                         | 71kB 17.2MB/s eta 0:00:01\r\u001b[K     |███████▍                        | 81kB 18.4MB/s eta 0:00:01\r\u001b[K     |████████▎                       | 92kB 17.5MB/s eta 0:00:01\r\u001b[K     |█████████▏                      | 102kB 17.2MB/s eta 0:00:01\r\u001b[K     |██████████▏                     | 112kB 17.2MB/s eta 0:00:01\r\u001b[K     |███████████                     | 122kB 17.2MB/s eta 0:00:01\r\u001b[K     |████████████                    | 133kB 17.2MB/s eta 0:00:01\r\u001b[K     |████████████▉                   | 143kB 17.2MB/s eta 0:00:01\r\u001b[K     |█████████████▉                  | 153kB 17.2MB/s eta 0:00:01\r\u001b[K     |██████████████▊                 | 163kB 17.2MB/s eta 0:00:01\r\u001b[K     |███████████████▋                | 174kB 17.2MB/s eta 0:00:01\r\u001b[K     |████████████████▋               | 184kB 17.2MB/s eta 0:00:01\r\u001b[K     |█████████████████▌              | 194kB 17.2MB/s eta 0:00:01\r\u001b[K     |██████████████████▍             | 204kB 17.2MB/s eta 0:00:01\r\u001b[K     |███████████████████▎            | 215kB 17.2MB/s eta 0:00:01\r\u001b[K     |████████████████████▎           | 225kB 17.2MB/s eta 0:00:01\r\u001b[K     |█████████████████████▏          | 235kB 17.2MB/s eta 0:00:01\r\u001b[K     |██████████████████████          | 245kB 17.2MB/s eta 0:00:01\r\u001b[K     |███████████████████████         | 256kB 17.2MB/s eta 0:00:01\r\u001b[K     |████████████████████████        | 266kB 17.2MB/s eta 0:00:01\r\u001b[K     |████████████████████████▉       | 276kB 17.2MB/s eta 0:00:01\r\u001b[K     |█████████████████████████▊      | 286kB 17.2MB/s eta 0:00:01\r\u001b[K     |██████████████████████████▊     | 296kB 17.2MB/s eta 0:00:01\r\u001b[K     |███████████████████████████▋    | 307kB 17.2MB/s eta 0:00:01\r\u001b[K     |████████████████████████████▌   | 317kB 17.2MB/s eta 0:00:01\r\u001b[K     |█████████████████████████████▍  | 327kB 17.2MB/s eta 0:00:01\r\u001b[K     |██████████████████████████████▍ | 337kB 17.2MB/s eta 0:00:01\r\u001b[K     |███████████████████████████████▎| 348kB 17.2MB/s eta 0:00:01\r\u001b[K     |████████████████████████████████| 358kB 17.2MB/s \n",
            "\u001b[?25hRequirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from shap) (1.19.5)\n",
            "Requirement already satisfied: scipy in /usr/local/lib/python3.7/dist-packages (from shap) (1.4.1)\n",
            "Requirement already satisfied: scikit-learn in /usr/local/lib/python3.7/dist-packages (from shap) (0.22.2.post1)\n",
            "Requirement already satisfied: pandas in /usr/local/lib/python3.7/dist-packages (from shap) (1.1.5)\n",
            "Requirement already satisfied: tqdm>4.25.0 in /usr/local/lib/python3.7/dist-packages (from shap) (4.41.1)\n",
            "Collecting slicer==0.0.7\n",
            "  Downloading https://files.pythonhosted.org/packages/78/c2/b3f55dfdb8af9812fdb9baf70cacf3b9e82e505b2bd4324d588888b81202/slicer-0.0.7-py3-none-any.whl\n",
            "Requirement already satisfied: numba in /usr/local/lib/python3.7/dist-packages (from shap) (0.51.2)\n",
            "Requirement already satisfied: cloudpickle in /usr/local/lib/python3.7/dist-packages (from shap) (1.3.0)\n",
            "Requirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.7/dist-packages (from scikit-learn->shap) (1.0.1)\n",
            "Requirement already satisfied: python-dateutil>=2.7.3 in /usr/local/lib/python3.7/dist-packages (from pandas->shap) (2.8.1)\n",
            "Requirement already satisfied: pytz>=2017.2 in /usr/local/lib/python3.7/dist-packages (from pandas->shap) (2018.9)\n",
            "Requirement already satisfied: llvmlite<0.35,>=0.34.0.dev0 in /usr/local/lib/python3.7/dist-packages (from numba->shap) (0.34.0)\n",
            "Requirement already satisfied: setuptools in /usr/local/lib/python3.7/dist-packages (from numba->shap) (56.0.0)\n",
            "Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.7/dist-packages (from python-dateutil>=2.7.3->pandas->shap) (1.15.0)\n",
            "Building wheels for collected packages: shap\n",
            "  Building wheel for shap (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for shap: filename=shap-0.39.0-cp37-cp37m-linux_x86_64.whl size=491632 sha256=5bf5c02d15c2e4b9b910b791186825699ef4b195278b4a98212574f836ece806\n",
            "  Stored in directory: /root/.cache/pip/wheels/15/27/f5/a8ab9da52fd159aae6477b5ede6eaaec69fd130fa0fa59f283\n",
            "Successfully built shap\n",
            "Installing collected packages: slicer, shap\n",
            "Successfully installed shap-0.39.0 slicer-0.0.7\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pUd6BVvsrtHX"
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import sklearn\n",
        "import shap"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dNs_YSsQxyxf"
      },
      "source": [
        "# Prepare Housing Dataset"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "omcfZwvetCsN"
      },
      "source": [
        "## Load data from the Colaboratory samples"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rPrEH0XrrxKL"
      },
      "source": [
        "housing_train_pd = pd.read_csv('/content/sample_data/california_housing_train.csv')\n",
        "housing_test_pd = pd.read_csv('/content/sample_data/california_housing_test.csv')"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gLfPSIB2sEhg",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 224
        },
        "outputId": "fae1b0b5-b516-42ec-df48-dc2a5712ba9d"
      },
      "source": [
        "housing_train_pd.head(5)"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>longitude</th>\n",
              "      <th>latitude</th>\n",
              "      <th>housing_median_age</th>\n",
              "      <th>total_rooms</th>\n",
              "      <th>total_bedrooms</th>\n",
              "      <th>population</th>\n",
              "      <th>households</th>\n",
              "      <th>median_income</th>\n",
              "      <th>median_house_value</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>-114.31</td>\n",
              "      <td>34.19</td>\n",
              "      <td>15.0</td>\n",
              "      <td>5612.0</td>\n",
              "      <td>1283.0</td>\n",
              "      <td>1015.0</td>\n",
              "      <td>472.0</td>\n",
              "      <td>1.4936</td>\n",
              "      <td>66900.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>-114.47</td>\n",
              "      <td>34.40</td>\n",
              "      <td>19.0</td>\n",
              "      <td>7650.0</td>\n",
              "      <td>1901.0</td>\n",
              "      <td>1129.0</td>\n",
              "      <td>463.0</td>\n",
              "      <td>1.8200</td>\n",
              "      <td>80100.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>-114.56</td>\n",
              "      <td>33.69</td>\n",
              "      <td>17.0</td>\n",
              "      <td>720.0</td>\n",
              "      <td>174.0</td>\n",
              "      <td>333.0</td>\n",
              "      <td>117.0</td>\n",
              "      <td>1.6509</td>\n",
              "      <td>85700.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>-114.57</td>\n",
              "      <td>33.64</td>\n",
              "      <td>14.0</td>\n",
              "      <td>1501.0</td>\n",
              "      <td>337.0</td>\n",
              "      <td>515.0</td>\n",
              "      <td>226.0</td>\n",
              "      <td>3.1917</td>\n",
              "      <td>73400.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>-114.57</td>\n",
              "      <td>33.57</td>\n",
              "      <td>20.0</td>\n",
              "      <td>1454.0</td>\n",
              "      <td>326.0</td>\n",
              "      <td>624.0</td>\n",
              "      <td>262.0</td>\n",
              "      <td>1.9250</td>\n",
              "      <td>65500.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   longitude  latitude  ...  median_income  median_house_value\n",
              "0    -114.31     34.19  ...         1.4936             66900.0\n",
              "1    -114.47     34.40  ...         1.8200             80100.0\n",
              "2    -114.56     33.69  ...         1.6509             85700.0\n",
              "3    -114.57     33.64  ...         3.1917             73400.0\n",
              "4    -114.57     33.57  ...         1.9250             65500.0\n",
              "\n",
              "[5 rows x 9 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OY_rbbattdjF",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 224
        },
        "outputId": "c6d173ed-b9d3-45b7-fd47-369b6a3f6c27"
      },
      "source": [
        "housing_test_pd.head(5)"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>longitude</th>\n",
              "      <th>latitude</th>\n",
              "      <th>housing_median_age</th>\n",
              "      <th>total_rooms</th>\n",
              "      <th>total_bedrooms</th>\n",
              "      <th>population</th>\n",
              "      <th>households</th>\n",
              "      <th>median_income</th>\n",
              "      <th>median_house_value</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>-122.05</td>\n",
              "      <td>37.37</td>\n",
              "      <td>27.0</td>\n",
              "      <td>3885.0</td>\n",
              "      <td>661.0</td>\n",
              "      <td>1537.0</td>\n",
              "      <td>606.0</td>\n",
              "      <td>6.6085</td>\n",
              "      <td>344700.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>-118.30</td>\n",
              "      <td>34.26</td>\n",
              "      <td>43.0</td>\n",
              "      <td>1510.0</td>\n",
              "      <td>310.0</td>\n",
              "      <td>809.0</td>\n",
              "      <td>277.0</td>\n",
              "      <td>3.5990</td>\n",
              "      <td>176500.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>-117.81</td>\n",
              "      <td>33.78</td>\n",
              "      <td>27.0</td>\n",
              "      <td>3589.0</td>\n",
              "      <td>507.0</td>\n",
              "      <td>1484.0</td>\n",
              "      <td>495.0</td>\n",
              "      <td>5.7934</td>\n",
              "      <td>270500.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>-118.36</td>\n",
              "      <td>33.82</td>\n",
              "      <td>28.0</td>\n",
              "      <td>67.0</td>\n",
              "      <td>15.0</td>\n",
              "      <td>49.0</td>\n",
              "      <td>11.0</td>\n",
              "      <td>6.1359</td>\n",
              "      <td>330000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>-119.67</td>\n",
              "      <td>36.33</td>\n",
              "      <td>19.0</td>\n",
              "      <td>1241.0</td>\n",
              "      <td>244.0</td>\n",
              "      <td>850.0</td>\n",
              "      <td>237.0</td>\n",
              "      <td>2.9375</td>\n",
              "      <td>81700.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   longitude  latitude  ...  median_income  median_house_value\n",
              "0    -122.05     37.37  ...         6.6085            344700.0\n",
              "1    -118.30     34.26  ...         3.5990            176500.0\n",
              "2    -117.81     33.78  ...         5.7934            270500.0\n",
              "3    -118.36     33.82  ...         6.1359            330000.0\n",
              "4    -119.67     36.33  ...         2.9375             81700.0\n",
              "\n",
              "[5 rows x 9 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MZaKv-ylswkM"
      },
      "source": [
        "## Check if any of the variables contain nan"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0yrgyulasV98",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "15618a09-921c-4ed5-896d-bd51c38c3dce"
      },
      "source": [
        "print(housing_train_pd.isnull().any())\n",
        "print(housing_test_pd.isnull().any())"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "longitude             False\n",
            "latitude              False\n",
            "housing_median_age    False\n",
            "total_rooms           False\n",
            "total_bedrooms        False\n",
            "population            False\n",
            "households            False\n",
            "median_income         False\n",
            "median_house_value    False\n",
            "dtype: bool\n",
            "longitude             False\n",
            "latitude              False\n",
            "housing_median_age    False\n",
            "total_rooms           False\n",
            "total_bedrooms        False\n",
            "population            False\n",
            "households            False\n",
            "median_income         False\n",
            "median_house_value    False\n",
            "dtype: bool\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GxytXkuuwP1d"
      },
      "source": [
        "## Shuffle the dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2mT52W2HuCqy"
      },
      "source": [
        "train_pd = housing_train_pd.sample(frac=1).reset_index(drop=True)\n",
        "test_pd = housing_test_pd.sample(frac=1).reset_index(drop=True)"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WUy0A-heuZkp",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 224
        },
        "outputId": "cd353cd4-8b1f-48db-a954-f3999c944057"
      },
      "source": [
        "train_pd.head(5)"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>longitude</th>\n",
              "      <th>latitude</th>\n",
              "      <th>housing_median_age</th>\n",
              "      <th>total_rooms</th>\n",
              "      <th>total_bedrooms</th>\n",
              "      <th>population</th>\n",
              "      <th>households</th>\n",
              "      <th>median_income</th>\n",
              "      <th>median_house_value</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>-120.80</td>\n",
              "      <td>37.52</td>\n",
              "      <td>13.0</td>\n",
              "      <td>2920.0</td>\n",
              "      <td>481.0</td>\n",
              "      <td>1602.0</td>\n",
              "      <td>490.0</td>\n",
              "      <td>3.9286</td>\n",
              "      <td>145800.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>-122.20</td>\n",
              "      <td>37.83</td>\n",
              "      <td>26.0</td>\n",
              "      <td>1589.0</td>\n",
              "      <td>223.0</td>\n",
              "      <td>542.0</td>\n",
              "      <td>211.0</td>\n",
              "      <td>8.4010</td>\n",
              "      <td>351200.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>-118.20</td>\n",
              "      <td>34.69</td>\n",
              "      <td>5.0</td>\n",
              "      <td>9076.0</td>\n",
              "      <td>1503.0</td>\n",
              "      <td>7694.0</td>\n",
              "      <td>1278.0</td>\n",
              "      <td>4.8750</td>\n",
              "      <td>163400.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>-117.14</td>\n",
              "      <td>33.07</td>\n",
              "      <td>12.0</td>\n",
              "      <td>9302.0</td>\n",
              "      <td>1603.0</td>\n",
              "      <td>4074.0</td>\n",
              "      <td>1504.0</td>\n",
              "      <td>4.3513</td>\n",
              "      <td>199600.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>-117.71</td>\n",
              "      <td>34.09</td>\n",
              "      <td>36.0</td>\n",
              "      <td>2637.0</td>\n",
              "      <td>476.0</td>\n",
              "      <td>1385.0</td>\n",
              "      <td>483.0</td>\n",
              "      <td>4.1739</td>\n",
              "      <td>158700.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   longitude  latitude  ...  median_income  median_house_value\n",
              "0    -120.80     37.52  ...         3.9286            145800.0\n",
              "1    -122.20     37.83  ...         8.4010            351200.0\n",
              "2    -118.20     34.69  ...         4.8750            163400.0\n",
              "3    -117.14     33.07  ...         4.3513            199600.0\n",
              "4    -117.71     34.09  ...         4.1739            158700.0\n",
              "\n",
              "[5 rows x 9 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9FfssdtQwUXR",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 224
        },
        "outputId": "97f45123-03a8-49a3-9f0e-542056e0bb67"
      },
      "source": [
        "test_pd.head(5)"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>longitude</th>\n",
              "      <th>latitude</th>\n",
              "      <th>housing_median_age</th>\n",
              "      <th>total_rooms</th>\n",
              "      <th>total_bedrooms</th>\n",
              "      <th>population</th>\n",
              "      <th>households</th>\n",
              "      <th>median_income</th>\n",
              "      <th>median_house_value</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>-117.64</td>\n",
              "      <td>34.09</td>\n",
              "      <td>34.0</td>\n",
              "      <td>2839.0</td>\n",
              "      <td>659.0</td>\n",
              "      <td>1822.0</td>\n",
              "      <td>631.0</td>\n",
              "      <td>3.0500</td>\n",
              "      <td>121300.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>-121.81</td>\n",
              "      <td>37.25</td>\n",
              "      <td>5.0</td>\n",
              "      <td>1975.0</td>\n",
              "      <td>520.0</td>\n",
              "      <td>861.0</td>\n",
              "      <td>440.0</td>\n",
              "      <td>4.4565</td>\n",
              "      <td>159000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>-118.26</td>\n",
              "      <td>34.07</td>\n",
              "      <td>30.0</td>\n",
              "      <td>929.0</td>\n",
              "      <td>238.0</td>\n",
              "      <td>763.0</td>\n",
              "      <td>214.0</td>\n",
              "      <td>2.5227</td>\n",
              "      <td>187500.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>-118.37</td>\n",
              "      <td>34.19</td>\n",
              "      <td>41.0</td>\n",
              "      <td>2924.0</td>\n",
              "      <td>867.0</td>\n",
              "      <td>2751.0</td>\n",
              "      <td>836.0</td>\n",
              "      <td>2.1000</td>\n",
              "      <td>171600.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>-122.49</td>\n",
              "      <td>37.69</td>\n",
              "      <td>35.0</td>\n",
              "      <td>2576.0</td>\n",
              "      <td>443.0</td>\n",
              "      <td>1273.0</td>\n",
              "      <td>433.0</td>\n",
              "      <td>4.7391</td>\n",
              "      <td>272800.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   longitude  latitude  ...  median_income  median_house_value\n",
              "0    -117.64     34.09  ...         3.0500            121300.0\n",
              "1    -121.81     37.25  ...         4.4565            159000.0\n",
              "2    -118.26     34.07  ...         2.5227            187500.0\n",
              "3    -118.37     34.19  ...         2.1000            171600.0\n",
              "4    -122.49     37.69  ...         4.7391            272800.0\n",
              "\n",
              "[5 rows x 9 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QdEvNja11LIj"
      },
      "source": [
        "## Scale the variables and split out train and test"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nJgRji3H4u2X"
      },
      "source": [
        "from sklearn.preprocessing import MinMaxScaler\n",
        "scaler = MinMaxScaler()\n",
        "# Training\n",
        "X_train = scaler.fit_transform(train_pd.iloc[:, :-1])\n",
        "y_train = np.array(train_pd.iloc[:, -1:]).ravel()\n",
        "# Test\n",
        "X_test = scaler.transform(test_pd.iloc[:, :-1])\n",
        "y_test = np.array(test_pd.iloc[:, -1:]).ravel()"
      ],
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YQTjgqp3xdbv",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "6a42d1ca-e055-4c2d-a66a-8ce8641596c5"
      },
      "source": [
        "X_train"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[0.35358566, 0.52922423, 0.23529412, ..., 0.04481628, 0.08041441,\n",
              "        0.23645881],\n",
              "       [0.21414343, 0.56216791, 0.49019608, ..., 0.01510693, 0.03453379,\n",
              "        0.54489593],\n",
              "       [0.6125498 , 0.22848034, 0.07843137, ..., 0.21556097, 0.20999836,\n",
              "        0.30172687],\n",
              "       ...,\n",
              "       [0.59561753, 0.17003188, 0.39215686, ..., 0.05356092, 0.17974017,\n",
              "        0.22749341],\n",
              "       [0.30776892, 0.5674814 , 0.80392157, ..., 0.00709101, 0.01512909,\n",
              "        0.18427332],\n",
              "       [0.58964143, 0.17959617, 0.66666667, ..., 0.03595953, 0.06314751,\n",
              "        0.27194797]])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eyJobpfBxSSn",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5978456a-0bdb-4d46-c18b-fc1d4a57307f"
      },
      "source": [
        "y_train"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([145800., 351200., 163400., ..., 367700., 158300., 181500.])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bJjV3D8Rxizx",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "4ed60c9a-a198-4ca7-c21d-ed9ed6471b0a"
      },
      "source": [
        "X_test"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[0.66832669, 0.16471838, 0.64705882, ..., 0.05098237, 0.10360138,\n",
              "        0.17586654],\n",
              "       [0.25298805, 0.50053135, 0.07843137, ..., 0.02404776, 0.07219207,\n",
              "        0.2728652 ],\n",
              "       [0.60657371, 0.16259299, 0.56862745, ..., 0.02130105, 0.03502713,\n",
              "        0.13950152],\n",
              "       ...,\n",
              "       [0.2938247 , 0.64505845, 0.70588235, ..., 0.02183357, 0.05821411,\n",
              "        0.2516448 ],\n",
              "       [0.24900398, 0.76514346, 0.74509804, ..., 0.01737717, 0.04324947,\n",
              "        0.12299141],\n",
              "       [0.52788845, 0.29755579, 0.25490196, ..., 0.05409344, 0.14981089,\n",
              "        0.14922553]])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6XaW6JfBxiz3",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "6abf3b01-dc01-4355-a033-410a4b9a654d"
      },
      "source": [
        "y_test"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([121300., 159000., 187500., ..., 136400.,  85800., 175000.])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rebbMly36wHu"
      },
      "source": [
        "# Train and interpret models\n",
        "Currently implemented regressors:\n",
        "1.   XGBoost\n",
        "2.   Light Gradient Boost\n",
        "3.   Random Forest\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cFgBq3kgVnZe"
      },
      "source": [
        "### Shared Code"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "C6rxXrGc6yFC"
      },
      "source": [
        "from sklearn.metrics import r2_score, mean_squared_error\n",
        "from sklearn.model_selection import RandomizedSearchCV"
      ],
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "A9tyZ7qz7qQr"
      },
      "source": [
        "def evaluate_model(model_choice, X_train, y_train, X_test, y_test):\n",
        "  \"\"\"\n",
        "  Function to evaluate an sklearn model and print various statistics\n",
        "\n",
        "  Parameters:\n",
        "  model_choice (sklearn.model): Trained model that you are evaluating\n",
        "  X_train (numpy.ndarray): Training data\n",
        "  y_train (numpy.ndarray): Test labels\n",
        "  X_train (numpy.ndarray): Testing data\n",
        "  y_test (numpy.ndarray): Test labels\n",
        "\n",
        "  Returns: None\n",
        "  \"\"\"\n",
        "  y_pred = model_choice.predict(X_test)\n",
        "\n",
        "  print(\"Model Score for Training data: {}\".format(model_choice.score(X_train,y_train)))\n",
        "  print(\"Model Score for Testing data: {}\".format(r2_score(y_test,y_pred)))\n",
        "  print(\"Root Mean Squared Error is {}\".format(np.sqrt(mean_squared_error(y_test,y_pred))))"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "m4TfPkQnHwXJ"
      },
      "source": [
        "def calc_shap_values(model, X_source, bg_size=0):\n",
        "    \"\"\"\n",
        "    Calculates the shapley values for a given model's parameters\n",
        "\n",
        "    Allows an approximation of shapley values for a given model based on a number of samples from a data source\n",
        "\n",
        "    Parameters:\n",
        "    model (sklearn.model): Model compatible with the sklearn API\n",
        "    X_source (numpy.ndarray): Source ndarray to sample\n",
        "    bg_size (int): Number of samples to build the background data, default=0 to use all data\n",
        "\n",
        "    Returns:\n",
        "    model_shap_values: Shapley values for each prediction\n",
        "    X_shap: Dataframe used as background data\n",
        "    model_Explainer: shap.TreeExplainer created from model and X_shap\n",
        "    \"\"\"\n",
        "    # Passing bg_size = 0 means the data exhaustively sampled\n",
        "    if bg_size == 0:\n",
        "        X_shap = X_source\n",
        "    else:\n",
        "    # Reduce the background data\n",
        "        X_shap = shap.sample(X_source, bg_size)\n",
        "    \n",
        "    model_explainer = shap.TreeExplainer(model, X_shap)\n",
        "    # Produce the shapley values\n",
        "    model_shap_values = model_explainer.shap_values(X_shap, check_additivity=False)\n",
        "    return model_shap_values, X_shap, model_explainer"
      ],
      "execution_count": 56,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lHM-mQR4PifC"
      },
      "source": [
        "def ABS_SHAP(df_shap, df):\n",
        "  \"\"\"\n",
        "  Shows a simplified summary plot with colours indicating correlation to target variable\n",
        "\n",
        "  It highlights the correlations in colors. The red color means a feature is positively correlated with the target variable\n",
        "  Source: https://towardsdatascience.com/explain-your-model-with-the-shap-values-bc36aac4de3d\n",
        "\n",
        "  Parameters:\n",
        "  df_shap (numpy.ndarray): Array of shapley values\n",
        "  df (pandas.core.frame.DataFrame): Pandas dataframe of source data\n",
        "\n",
        "  Returns: matplotlib.axes._subplots.AxesSubplot\n",
        "  \"\"\"\n",
        "  #import matplotlib as plt\n",
        "  # Make a copy of the input data\n",
        "  shap_v = pd.DataFrame(df_shap)\n",
        "  feature_list = df.columns\n",
        "  shap_v.columns = feature_list\n",
        "  df_v = df.copy().reset_index().drop('index',axis=1)\n",
        "  \n",
        "  # Determine the correlation in order to plot with different colors\n",
        "  corr_list = list()\n",
        "  for i in feature_list:\n",
        "      b = np.corrcoef(shap_v[i],df_v[i])[1][0]\n",
        "      corr_list.append(b)\n",
        "  corr_df = pd.concat([pd.Series(feature_list),pd.Series(corr_list)],axis=1).fillna(0)\n",
        "  # Make a data frame. Column 1 is the feature, and Column 2 is the correlation coefficient\n",
        "  corr_df.columns  = ['Variable','Corr']\n",
        "  corr_df['Sign'] = np.where(corr_df['Corr']>0,'red','blue')\n",
        "  \n",
        "  # Plot it\n",
        "  shap_abs = np.abs(shap_v)\n",
        "  k=pd.DataFrame(shap_abs.mean()).reset_index()\n",
        "  k.columns = ['Variable','SHAP_abs']\n",
        "  k2 = k.merge(corr_df,left_on = 'Variable',right_on='Variable',how='inner')\n",
        "  k2 = k2.sort_values(by='SHAP_abs',ascending = True)\n",
        "  colorlist = k2['Sign']\n",
        "  ax = k2.plot.barh(x='Variable',y='SHAP_abs',color = colorlist, figsize=(5,6),legend=False)\n",
        "  ax.set_xlabel(\"SHAP Value (Red = Positive Impact)\")"
      ],
      "execution_count": 18,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1myKgmk6Fp5A"
      },
      "source": [
        "## XGBoost Regressor"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kD81vWS6I064"
      },
      "source": [
        "### Random search optimiser"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lFWM6bQNI069"
      },
      "source": [
        "# random_grid = {\n",
        "#         'num_boost_round': [10, 15, 20, 25, 30, 35, 40, 45, 50],\n",
        "#         'eta': [0.05, 0.1, 0.2, 0.3, 0.4, 0.5],\n",
        "#         'max_depth': [3, 4, 5, 6, 7, 8, 9],\n",
        "#         'subsample': np.arange(0.5, 1.0, 0.1),\n",
        "#         'colsample_bytree': np.arange(0.7, 1.0, 0.1),\n",
        "#         'n_estimators': [10, 20, 30, 40, 50],\n",
        "#         'alpha': [10, 20, 30, 40, 50]\n",
        "#     }"
      ],
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "o6QJi-xHI07F"
      },
      "source": [
        "# from xgboost import XGBRegressor\n",
        "# # Use the random grid to search for best hyperparameters\n",
        "# # First create the base model to tune\n",
        "# xg = XGBRegressor(tree_method='gpu_hist', gpu_id=0, objective ='reg:squarederror', learning_rate = 0.1)\n",
        "# # Random search of parameters, using 3 fold cross validation, \n",
        "# # search across 100 different combinations, and use all available cores\n",
        "# random_xg = RandomizedSearchCV(estimator = xg, param_distributions = random_grid, n_iter = 1000, cv = 3, verbose=2, random_state=42, n_jobs = -1)\n",
        "# # Fit the random search model\n",
        "# random_xg.fit(X_train, y_train)"
      ],
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "99slh0dAI07H"
      },
      "source": [
        "# random_xg.best_estimator_"
      ],
      "execution_count": 21,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3MZbnV1cI07J"
      },
      "source": [
        "# evaluate_model(random_xg, X_train, y_train, X_test, y_test)"
      ],
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bGq6h668K5hn"
      },
      "source": [
        "### Train an optimised XGBoost model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xsbBqy8FFsr1"
      },
      "source": [
        "# BUG IN LIBRARY PREVENTS GPU TRAINED XGBOOST FROM FUNCTIONING\n",
        "# Solution still needs to be merged into main code as of 9th May 2020\n",
        "\n",
        "# from xgboost import XGBRegressor\n",
        "# model_xg = XGBRegressor(alpha=10, base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
        "#              colsample_bynode=1, colsample_bytree=0.8999999999999999, eta=0.2,\n",
        "#              gamma=0, gpu_id=0, importance_type='gain', learning_rate=0.1,\n",
        "#              max_delta_step=0, max_depth=9, min_child_weight=1, missing=None,\n",
        "#              n_estimators=50, n_jobs=1, nthread=None, num_boost_round=50,\n",
        "#              objective='reg:squarederror', random_state=0, reg_alpha=0,\n",
        "#              reg_lambda=1, scale_pos_weight=1, seed=None, silent=None,\n",
        "#              subsample=0.8999999999999999, tree_method='gpu_hist', verbosity=1)\n",
        "# %time model_xg.fit(X_train, y_train)"
      ],
      "execution_count": 23,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7BnVnqxpMgRr",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "cf2578f5-e968-448d-ff85-c5b4677ff8e2"
      },
      "source": [
        "from xgboost import XGBRegressor\n",
        "# model_xg = XGBRegressor(alpha=20, base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
        "#              colsample_bynode=1, colsample_bytree=0.9, eta=0.3, gamma=0,\n",
        "#              importance_type='gain', learning_rate=0.1,\n",
        "#              max_delta_step=0, max_depth=5, min_child_weight=1, missing=None,\n",
        "#              n_estimators=30, n_jobs=1, nthread=None, num_boost_round=50,\n",
        "#              objective='reg:squarederror', random_state=0, reg_alpha=0,\n",
        "#              reg_lambda=1, scale_pos_weight=1, seed=None, silent=None,\n",
        "#              subsample=1.0, verbosity=1)\n",
        "\n",
        "model_xg = XGBRegressor(alpha=10, base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
        "             colsample_bynode=1, colsample_bytree=0.8999999999999999, eta=0.2,\n",
        "             gamma=0, importance_type='gain', learning_rate=0.1,\n",
        "             max_delta_step=0, max_depth=9, min_child_weight=1, missing=None,\n",
        "             n_estimators=50, n_jobs=1, nthread=None, num_boost_round=50,\n",
        "             objective='reg:squarederror', random_state=0, reg_alpha=0,\n",
        "             reg_lambda=1, scale_pos_weight=1, seed=None, silent=None,\n",
        "             subsample=0.8999999999999999, verbosity=1)\n",
        "\n",
        "%time model_xg.fit(X_train, y_train)"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 1.37 s, sys: 155 ms, total: 1.52 s\n",
            "Wall time: 2.2 s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "XGBRegressor(alpha=10, base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
              "             colsample_bynode=1, colsample_bytree=0.8999999999999999, eta=0.2,\n",
              "             gamma=0, importance_type='gain', learning_rate=0.1,\n",
              "             max_delta_step=0, max_depth=9, min_child_weight=1, missing=None,\n",
              "             n_estimators=50, n_jobs=1, nthread=None, num_boost_round=50,\n",
              "             objective='reg:squarederror', random_state=0, reg_alpha=0,\n",
              "             reg_lambda=1, scale_pos_weight=1, seed=None, silent=None,\n",
              "             subsample=0.8999999999999999, verbosity=1)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YhIonkLQGO0N",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "8232eecf-7a17-42ce-c926-43629e008fca"
      },
      "source": [
        "evaluate_model(model_xg, X_train, y_train, X_test, y_test)"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model Score for Training data: 0.9383415615059759\n",
            "Model Score for Testing data: 0.8210702676769799\n",
            "Root Mean Squared Error is 47841.750100383746\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MB9k0JqYyMRl"
      },
      "source": [
        "### Standard feature importance plots\n",
        "Here we use the standard feature importance plots common to sklearn and xgboost libraries.\n",
        "\n",
        "The key takeaway is that these built in plots can show conflicting, disparate or ambiguous results."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YR9BGHDbBnnJ"
      },
      "source": [
        "import xgboost"
      ],
      "execution_count": 26,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Aier6VlzyRc1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "outputId": "805339c2-671b-4d13-ac2a-93604407bf0e"
      },
      "source": [
        "xgboost.plot_importance(model_xg).set_yticklabels(test_pd.columns[:-1])"
      ],
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[Text(0, 0, 'longitude'),\n",
              " Text(0, 0, 'latitude'),\n",
              " Text(0, 0, 'housing_median_age'),\n",
              " Text(0, 0, 'total_rooms'),\n",
              " Text(0, 0, 'total_bedrooms'),\n",
              " Text(0, 0, 'population'),\n",
              " Text(0, 0, 'households'),\n",
              " Text(0, 0, 'median_income')]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 27
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OKkWh3woyaEO",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "outputId": "93cbd3c1-c639-4a1d-ec60-797dc6d19d03"
      },
      "source": [
        "xgboost.plot_importance(model_xg, importance_type=\"cover\").set_yticklabels(test_pd.columns[:-1])"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[Text(0, 0, 'longitude'),\n",
              " Text(0, 0, 'latitude'),\n",
              " Text(0, 0, 'housing_median_age'),\n",
              " Text(0, 0, 'total_rooms'),\n",
              " Text(0, 0, 'total_bedrooms'),\n",
              " Text(0, 0, 'population'),\n",
              " Text(0, 0, 'households'),\n",
              " Text(0, 0, 'median_income')]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uzuwkimPyaLI",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "outputId": "2d55fb28-76b7-4bbf-8ca1-0dad89d59163"
      },
      "source": [
        "xgboost.plot_importance(model_xg, importance_type=\"gain\").set_yticklabels(test_pd.columns[:-1])"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[Text(0, 0, 'longitude'),\n",
              " Text(0, 0, 'latitude'),\n",
              " Text(0, 0, 'housing_median_age'),\n",
              " Text(0, 0, 'total_rooms'),\n",
              " Text(0, 0, 'total_bedrooms'),\n",
              " Text(0, 0, 'population'),\n",
              " Text(0, 0, 'households'),\n",
              " Text(0, 0, 'median_income')]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4mI6aU2rL5oL"
      },
      "source": [
        "### Use Shap library to explain feature contribution\n",
        "We can use the Shap library to better explain feature contribution in terms of both magnitude, specific value impact and positive/negative contribution."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FYiOLaWhFrwW",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "8771066b-52e3-4c12-89e1-b36defd9dc50"
      },
      "source": [
        "# Calculate shapley values for the model\n",
        "%time xg_shap_values, X_shap, _ = calc_shap_values(model_xg, X_source=X_test, bg_size=1000)"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 6.49 s, sys: 12.6 ms, total: 6.5 s\n",
            "Wall time: 6.48 s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "l-JqrSuQ63Zj"
      },
      "source": [
        "The first plot is a variable importance plot lists the most significant variables in descending order.\n",
        "\n",
        "The top variables contribute more to the model than the bottom ones and thus have high predictive power."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ECH8eGo2L5oS",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 320
        },
        "outputId": "b5b5180b-28e0-485f-9c8e-111821a9a31f"
      },
      "source": [
        "shap.summary_plot(xg_shap_values, X_shap, plot_type=\"bar\", feature_names=test_pd.columns[:-1])"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x338.4 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qUgUKZwl56I6"
      },
      "source": [
        "The second plot is made of all the dots in the testing data. It demonstrates the following information:\n",
        "1. Feature importance: Variables are ranked in descending order.\n",
        "2. Impact: The horizontal location shows whether the effect of that value is associated with a higher or lower prediction.\n",
        "3. Original value: Color shows whether that variable is high (in red) or low (in blue) for that observation.\n",
        "4. Correlation: A high level of the “median_income” content has a high and positive impact on the price estimation. The “high” comes from the red color, and the “positive” impact is shown on the X-axis. Similarly, we will say the “latitude” is negatively correlated with the target variable."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MmLoI2HPM3bW",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 324
        },
        "outputId": "4eba0a97-5212-4fdd-aeef-793636a83df2"
      },
      "source": [
        "shap.summary_plot(xg_shap_values, X_shap, feature_names=test_pd.columns[:-1])"
      ],
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x338.4 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uIoXGeJU8N7L"
      },
      "source": [
        "The final plot shows a simplified summary plot with colours indicating correlation to target variable\n",
        "\n",
        "  It highlights the correlations in colors. The red color means a feature is positively correlated with the target variable\n",
        "  Source: https://towardsdatascience.com/explain-your-model-with-the-shap-values-bc36aac4de3d"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7A0yos3iNW3X",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 388
        },
        "outputId": "0d8a2920-66ab-4406-ee14-112e8588ef1e"
      },
      "source": [
        "ABS_SHAP(xg_shap_values, pd.DataFrame(X_shap, columns=test_pd.columns[:-1])) "
      ],
      "execution_count": 33,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 360x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1Ol_4_qYG9PH"
      },
      "source": [
        "## Light Gradient Boost Regressor"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NLuUsAHDQsob"
      },
      "source": [
        "### Random search optimiser"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rZ866Ch9Qsoc"
      },
      "source": [
        "# random_grid = {\n",
        "#     'learning_rate': [0.05,0.1],\n",
        "#     'num_leaves': [10,20,30,40,90,200],\n",
        "#     'boosting_type' : ['gbdt'],\n",
        "#     'objective' : ['binary'],\n",
        "#     'max_depth' : [5,6,7,8,-1],\n",
        "#     'random_state' : [501,None], \n",
        "#     'colsample_bytree' : [0.5,0.7,1.0],\n",
        "#     'subsample' : [0.5,0.7,1.0],\n",
        "#     'min_split_gain' : [0.0,0.01],\n",
        "#     'min_data_in_leaf':[10],\n",
        "#     'metric':['auc']\n",
        "#     }"
      ],
      "execution_count": 34,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DTKYlekkQsof"
      },
      "source": [
        "# from lightgbm import LGBMRegressor\n",
        "\n",
        "# # Use the random grid to search for best hyperparameters\n",
        "# # First create the base model to tune\n",
        "# lg = LGBMRegressor()\n",
        "# # Random search of parameters, using 3 fold cross validation, \n",
        "# # search across 100 different combinations, and use all available cores\n",
        "# random_lg = RandomizedSearchCV(estimator = lg, param_distributions = random_grid, n_iter = 1000, cv = 3, verbose=2, random_state=42, n_jobs = -1)\n",
        "# # Fit the random search model\n",
        "# random_lg.fit(X_train, y_train)"
      ],
      "execution_count": 35,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JqU2hZUBQsoi"
      },
      "source": [
        "# random_lg.best_estimator_"
      ],
      "execution_count": 36,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UvEFy5Z1Qsol"
      },
      "source": [
        "# evaluate_model(random_lg, X_train, y_train, X_test, y_test)"
      ],
      "execution_count": 37,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qvIU2frGQsor"
      },
      "source": [
        "### Train an optimised LGBoost model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NgfLlVl3PsYQ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "54b12dcb-4e58-4f9d-a197-5d745318be13"
      },
      "source": [
        "from lightgbm import LGBMRegressor\n",
        "model_lg = LGBMRegressor()\n",
        "%time model_lg.fit(X_train, y_train)"
      ],
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 261 ms, sys: 6.7 ms, total: 268 ms\n",
            "Wall time: 282 ms\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "LGBMRegressor(boosting_type='gbdt', class_weight=None, colsample_bytree=1.0,\n",
              "              importance_type='split', learning_rate=0.1, max_depth=-1,\n",
              "              min_child_samples=20, min_child_weight=0.001, min_split_gain=0.0,\n",
              "              n_estimators=100, n_jobs=-1, num_leaves=31, objective=None,\n",
              "              random_state=None, reg_alpha=0.0, reg_lambda=0.0, silent=True,\n",
              "              subsample=1.0, subsample_for_bin=200000, subsample_freq=0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 38
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EIzF92gUQe25",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "6c44a1f1-53f3-4aa8-a6a8-283687ef8906"
      },
      "source": [
        "evaluate_model(model_lg, X_train, y_train, X_test, y_test)"
      ],
      "execution_count": 39,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model Score for Training data: 0.8750530663252905\n",
            "Model Score for Testing data: 0.8163254105424506\n",
            "Root Mean Squared Error is 48471.93320137177\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PpKK0dsTc5_e"
      },
      "source": [
        "### Standard feature importance plots\n",
        "Here we use the standard feature importance plots common to sklearn and xgboost libraries.\n",
        "\n",
        "The key takeaway is that these built in plots can show conflicting, disparate or ambiguous results."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bJd5DEqDc5_j"
      },
      "source": [
        "import lightgbm"
      ],
      "execution_count": 40,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bSO4BxCPc5_0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "outputId": "be3657b7-daea-4101-81be-1f623d55d848"
      },
      "source": [
        "lightgbm.plot_importance(model_lg).set_yticklabels(test_pd.columns[:-1])"
      ],
      "execution_count": 41,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[Text(0, 0, 'longitude'),\n",
              " Text(0, 0, 'latitude'),\n",
              " Text(0, 0, 'housing_median_age'),\n",
              " Text(0, 0, 'total_rooms'),\n",
              " Text(0, 0, 'total_bedrooms'),\n",
              " Text(0, 0, 'population'),\n",
              " Text(0, 0, 'households'),\n",
              " Text(0, 0, 'median_income')]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 41
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zp37LBFrc6AQ",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "outputId": "ed0662b2-b338-485b-9f09-6b06743e310c"
      },
      "source": [
        "lightgbm.plot_importance(model_lg, importance_type=\"gain\").set_yticklabels(test_pd.columns[:-1])"
      ],
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[Text(0, 0, 'longitude'),\n",
              " Text(0, 0, 'latitude'),\n",
              " Text(0, 0, 'housing_median_age'),\n",
              " Text(0, 0, 'total_rooms'),\n",
              " Text(0, 0, 'total_bedrooms'),\n",
              " Text(0, 0, 'population'),\n",
              " Text(0, 0, 'households'),\n",
              " Text(0, 0, 'median_income')]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 42
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Nw797Q7cS-gQ"
      },
      "source": [
        "### Use Shap library to explain feature contribution\n",
        "BUG: The aditivity check fails in the TreeExplainer on LGBoost \n",
        "https://github.com/slundberg/shap/issues/1071\n",
        "\n",
        "Adjust the bg_size downwards to fix this\n",
        "* 600 seems to be the limit as of 10 May 2020"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AEyAeoKHS-gW",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "bd65e9ac-c699-4398-9a14-98ed85c364d4"
      },
      "source": [
        "# Calculate shapley values for the model\n",
        "%time lg_shap_values, X_shap, _ = calc_shap_values(model_lg, X_source=X_test, bg_size=180)"
      ],
      "execution_count": 43,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 785 ms, sys: 3.67 ms, total: 789 ms\n",
            "Wall time: 799 ms\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FfAgOqMmS-gY",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 320
        },
        "outputId": "ceb7d07e-58f3-4df8-d39c-b6af30cc10e6"
      },
      "source": [
        "shap.summary_plot(lg_shap_values, X_test, plot_type=\"bar\", feature_names=test_pd.columns[:-1])"
      ],
      "execution_count": 44,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x338.4 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yjlplb5OS-gb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 324
        },
        "outputId": "2cc6f9bd-f5de-4eb3-c021-1d7563d9db01"
      },
      "source": [
        "shap.summary_plot(lg_shap_values, X_shap, feature_names=test_pd.columns[:-1])"
      ],
      "execution_count": 45,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x338.4 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "E9NbmSbaS-gd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 388
        },
        "outputId": "b923b21d-5747-4bce-ba77-415254b1d550"
      },
      "source": [
        "ABS_SHAP(lg_shap_values, pd.DataFrame(X_shap, columns=test_pd.columns[:-1])) "
      ],
      "execution_count": 46,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 360x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "S0ugqimG6-Yd"
      },
      "source": [
        "## Random Forest Regression"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Z6tp_qgFzY75"
      },
      "source": [
        "### Random search optimiser"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2pMvoPa19LLL"
      },
      "source": [
        "# from sklearn.model_selection import RandomizedSearchCV\n",
        "# # Number of trees in random forest\n",
        "# n_estimators = [int(x) for x in np.linspace(start = 200, stop = 2000, num = 10)]\n",
        "# # Number of features to consider at every split\n",
        "# max_features = ['auto', 'sqrt']\n",
        "# # Maximum number of levels in tree\n",
        "# max_depth = [int(x) for x in np.linspace(10, 110, num = 11)]\n",
        "# max_depth.append(None)\n",
        "# # Minimum number of samples required to split a node\n",
        "# min_samples_split = [2, 5, 10]\n",
        "# # Minimum number of samples required at each leaf node\n",
        "# min_samples_leaf = [1, 2, 4]\n",
        "# # Method of selecting samples for training each tree\n",
        "# bootstrap = [True, False]\n",
        "# # Create the random grid\n",
        "# random_grid = {'n_estimators': n_estimators,\n",
        "#                'max_features': max_features,\n",
        "#                'max_depth': max_depth,\n",
        "#                'min_samples_split': min_samples_split,\n",
        "#                'min_samples_leaf': min_samples_leaf,\n",
        "#                'bootstrap': bootstrap}\n",
        "# print(random_grid)"
      ],
      "execution_count": 47,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ockzq03x9Tn7"
      },
      "source": [
        "# from sklearn.ensemble import RandomForestRegressor\n",
        "# # Use the random grid to search for best hyperparameters\n",
        "# # First create the base model to tune\n",
        "# rf = RandomForestRegressor(n_jobs=-1)\n",
        "# # Random search of parameters, using 3 fold cross validation, \n",
        "# # search across 100 different combinations, and use all available cores\n",
        "# random_rf = RandomizedSearchCV(estimator = rf, param_distributions = random_grid, n_iter = 100, cv = 3, verbose=2, random_state=42, n_jobs = -1)\n",
        "# # Fit the random search model\n",
        "# random_rf.fit(X_train, y_train)"
      ],
      "execution_count": 48,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lNfLkiJGyxWM"
      },
      "source": [
        "# random_rf.best_estimator_"
      ],
      "execution_count": 49,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DBlRFmds9q5W"
      },
      "source": [
        "# evaluate_model(random_rf, X_train, y_train, X_test, y_test)"
      ],
      "execution_count": 50,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SvLRR7Kczh3V"
      },
      "source": [
        "### Train optimised Random Forest model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sQdmFZJk7Dsb",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "46d6dd63-99f5-4cb7-db2d-719a5521a5f5"
      },
      "source": [
        "from sklearn.ensemble import RandomForestRegressor\n",
        "model_rf = RandomForestRegressor(bootstrap=True, ccp_alpha=0.0, criterion='mse',\n",
        "                      max_depth=80, max_features='auto', max_leaf_nodes=None,\n",
        "                      max_samples=None, min_impurity_decrease=0.0,\n",
        "                      min_impurity_split=None, min_samples_leaf=2,\n",
        "                      min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
        "                      n_estimators=1800, n_jobs=-1, oob_score=False,\n",
        "                      random_state=None, verbose=0, warm_start=False)\n",
        "%time model_rf.fit(X_train,y_train)"
      ],
      "execution_count": 51,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 3min 52s, sys: 1.06 s, total: 3min 53s\n",
            "Wall time: 1min 59s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "RandomForestRegressor(bootstrap=True, ccp_alpha=0.0, criterion='mse',\n",
              "                      max_depth=80, max_features='auto', max_leaf_nodes=None,\n",
              "                      max_samples=None, min_impurity_decrease=0.0,\n",
              "                      min_impurity_split=None, min_samples_leaf=2,\n",
              "                      min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
              "                      n_estimators=1800, n_jobs=-1, oob_score=False,\n",
              "                      random_state=None, verbose=0, warm_start=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 51
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3YqarU5I8BJ8",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "006030e9-f7ea-4c3f-a575-3a11623184d6"
      },
      "source": [
        "evaluate_model(model_rf, X_train, y_train, X_test, y_test)"
      ],
      "execution_count": 52,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model Score for Training data: 0.9605185433296777\n",
            "Model Score for Testing data: 0.8102397747514685\n",
            "Root Mean Squared Error is 49268.392850579774\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ojJ68EqJdQqn"
      },
      "source": [
        "### Standard feature importance plots\n",
        "Here we use the standard feature importance plots common to sklearn and xgboost libraries.\n",
        "\n",
        "The key takeaway is that these built in plots can show conflicting, disparate or ambiguous results."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ufSUfuN8dhIj",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "outputId": "b5946f0e-0515-4a00-a0c8-4aec3cc81a02"
      },
      "source": [
        "feat_importances = pd.Series(model_rf.feature_importances_, index=test_pd.columns[:-1])\n",
        "feat_importances.nlargest(len(test_pd.columns[:-1]))[::-1].plot(kind='barh')"
      ],
      "execution_count": 53,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7f385baa2850>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 53
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OXcJ7BMU2otf"
      },
      "source": [
        "### Use Shap library to explain feature contribution"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0iw_jhDCZPOw",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b8d465f5-c4b6-49b1-863b-646dfe9cfc76"
      },
      "source": [
        "# Calculate shapley values for the model\n",
        "%time rf_shap_values, X_shap, _ = calc_shap_values(model_rf, X_source=X_test, bg_size=100)"
      ],
      "execution_count": 57,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            " 98%|===================| 98/100 [01:29<00:01]       "
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 1min 30s, sys: 817 ms, total: 1min 31s\n",
            "Wall time: 1min 31s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_-93Jf25ZPO5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 320
        },
        "outputId": "e24d1907-2489-4a90-8cc0-a10a51814f4e"
      },
      "source": [
        "shap.summary_plot(rf_shap_values, X_test, plot_type=\"bar\", feature_names=test_pd.columns[:-1])"
      ],
      "execution_count": 58,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x338.4 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iO_mQ0VJZPO7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 324
        },
        "outputId": "d44a612d-3d6e-473e-a7f5-a8a43582b68a"
      },
      "source": [
        "shap.summary_plot(rf_shap_values, X_shap, feature_names=test_pd.columns[:-1])"
      ],
      "execution_count": 59,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x338.4 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6azJq1KeZPO-",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 388
        },
        "outputId": "44c88514-e74d-4651-9ae8-252822312ab4"
      },
      "source": [
        "ABS_SHAP(rf_shap_values, pd.DataFrame(X_shap, columns=test_pd.columns[:-1])) "
      ],
      "execution_count": 60,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 360x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LQcDzWx1Ptt5"
      },
      "source": [
        "# Unused Regression Models"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rEV_Vvk71_J5"
      },
      "source": [
        "## Train a Bayesian Ridge Regressor"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rFbmgdxf2CFG",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        },
        "outputId": "db9e8ce0-5217-45ba-c9cd-11363a6c2449"
      },
      "source": [
        "from sklearn.linear_model import BayesianRidge\n",
        "model_br = BayesianRidge()\n",
        "%time model_br.fit(X_train,y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 16.2 ms, sys: 14 ms, total: 30.2 ms\n",
            "Wall time: 24.7 ms\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "BayesianRidge(alpha_1=1e-06, alpha_2=1e-06, alpha_init=None,\n",
              "              compute_score=False, copy_X=True, fit_intercept=True,\n",
              "              lambda_1=1e-06, lambda_2=1e-06, lambda_init=None, n_iter=300,\n",
              "              normalize=False, tol=0.001, verbose=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 171
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "d0RfsE_l2VHP",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        },
        "outputId": "fe9fc42d-797b-49d3-b2a6-86dbd8550058"
      },
      "source": [
        "evaluate_model(model_br, X_train, y_train, X_test, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model Score for Training data: 0.6413373752225843\n",
            "Model Score for Testing data: 0.6195245662847987\n",
            "Root Mean Squared Error is 69763.63680959698\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2Vgeh3MN1D5c"
      },
      "source": [
        "## Train a Decision Tree Regressor"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cVXff3nD1IzZ",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 153
        },
        "outputId": "1d9f3ca0-851d-4611-85e1-90e0826600bb"
      },
      "source": [
        "from sklearn.tree import DecisionTreeRegressor\n",
        "model_dt = DecisionTreeRegressor()\n",
        "%time model_dt.fit(X_train,y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 217 ms, sys: 77 ms, total: 294 ms\n",
            "Wall time: 210 ms\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "DecisionTreeRegressor(ccp_alpha=0.0, criterion='mse', max_depth=None,\n",
              "                      max_features=None, max_leaf_nodes=None,\n",
              "                      min_impurity_decrease=0.0, min_impurity_split=None,\n",
              "                      min_samples_leaf=1, min_samples_split=2,\n",
              "                      min_weight_fraction_leaf=0.0, presort='deprecated',\n",
              "                      random_state=None, splitter='best')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 173
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JC1gkg-l1Zti",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        },
        "outputId": "001b60aa-37e4-4d85-ae2e-c43731fd1554"
      },
      "source": [
        "evaluate_model(model_dt, X_train, y_train, X_test, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model Score for Training data: 1.0\n",
            "Model Score for Testing data: 0.6490075367587544\n",
            "Root Mean Squared Error is 67006.15500216131\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1_uGalxU6bCz"
      },
      "source": [
        "## Train a Linear Regressor"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lAe0mGG76dWK",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        },
        "outputId": "4dc34f41-65dd-45eb-f7dd-0e34678eb29b"
      },
      "source": [
        "from sklearn.linear_model import LinearRegression\n",
        "model_lr = LinearRegression()\n",
        "%time model_lr.fit(X_train,y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 4.44 ms, sys: 5 ms, total: 9.44 ms\n",
            "Wall time: 9.45 ms\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "LinearRegression(copy_X=True, fit_intercept=True, n_jobs=None, normalize=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 175
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HEOc5OT46t2r",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        },
        "outputId": "1a2a4361-4d6b-46c4-d5f9-11774d19e856"
      },
      "source": [
        "evaluate_model(model_rf, X_train, y_train, X_test, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model Score for Training data: 0.9605582689976389\n",
            "Model Score for Testing data: 0.8108134243496904\n",
            "Root Mean Squared Error is 49193.86673256686\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "i0gTcGq0xs4a"
      },
      "source": [
        "## Train a Support Vector Regressor\n",
        "This techniques finds the optimal hyperplane to seperate observations into classes"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4MpO_oUFz8tG"
      },
      "source": [
        "from sklearn.svm import SVR"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BsBbzPKUxqEb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        },
        "outputId": "df559c19-6fdb-439e-b6ee-a5089857b878"
      },
      "source": [
        "# Support Vector Regression\n",
        "model_svr = SVR(kernel=\"rbf\", cache_size=32000)\n",
        "%time model_svr.fit(X_train,y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 19.2 s, sys: 172 ms, total: 19.4 s\n",
            "Wall time: 19.4 s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SVR(C=1.0, cache_size=32000, coef0=0.0, degree=3, epsilon=0.1, gamma='scale',\n",
              "    kernel='rbf', max_iter=-1, shrinking=True, tol=0.001, verbose=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 178
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dtalk4JHzx56",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        },
        "outputId": "6f19620e-c38b-4edd-db50-dd2ccac1505a"
      },
      "source": [
        "evaluate_model(model_rf, X_train, y_train, X_test, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model Score for Training data: 0.9605582689976389\n",
            "Model Score for Testing data: 0.8108134243496904\n",
            "Root Mean Squared Error is 49193.86673256686\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}