08-gradient-boosting-machines/5-xgboost-local.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "5e9e49bd",
   "metadata": {},
   "source": [
    "# xgboost local interpretation\n",
    "\n",
    "- [Online Course](https://www.trainindata.com/p/machine-learning-interpretability)\n",
    "\n",
    "Here I show how to make local interpretations for a regression model, but the code for classification is the same. We just need to change the model class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "b619bd11",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "from sklearn.datasets import fetch_california_housing\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "from xgboost import XGBRegressor"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "750e70b0",
   "metadata": {},
   "source": [
    "### Load data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "050acc05",
   "metadata": {},
   "outputs": [
    {
     "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>MedInc</th>\n",
       "      <th>HouseAge</th>\n",
       "      <th>AveRooms</th>\n",
       "      <th>AveBedrms</th>\n",
       "      <th>Population</th>\n",
       "      <th>AveOccup</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>8.3252</td>\n",
       "      <td>41.0</td>\n",
       "      <td>6.984127</td>\n",
       "      <td>1.023810</td>\n",
       "      <td>322.0</td>\n",
       "      <td>2.555556</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>8.3014</td>\n",
       "      <td>21.0</td>\n",
       "      <td>6.238137</td>\n",
       "      <td>0.971880</td>\n",
       "      <td>2401.0</td>\n",
       "      <td>2.109842</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>7.2574</td>\n",
       "      <td>52.0</td>\n",
       "      <td>8.288136</td>\n",
       "      <td>1.073446</td>\n",
       "      <td>496.0</td>\n",
       "      <td>2.802260</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>5.6431</td>\n",
       "      <td>52.0</td>\n",
       "      <td>5.817352</td>\n",
       "      <td>1.073059</td>\n",
       "      <td>558.0</td>\n",
       "      <td>2.547945</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3.8462</td>\n",
       "      <td>52.0</td>\n",
       "      <td>6.281853</td>\n",
       "      <td>1.081081</td>\n",
       "      <td>565.0</td>\n",
       "      <td>2.181467</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   MedInc  HouseAge  AveRooms  AveBedrms  Population  AveOccup\n",
       "0  8.3252      41.0  6.984127   1.023810       322.0  2.555556\n",
       "1  8.3014      21.0  6.238137   0.971880      2401.0  2.109842\n",
       "2  7.2574      52.0  8.288136   1.073446       496.0  2.802260\n",
       "3  5.6431      52.0  5.817352   1.073059       558.0  2.547945\n",
       "4  3.8462      52.0  6.281853   1.081081       565.0  2.181467"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# load the California House price data from Scikit-learn\n",
    "X, y = fetch_california_housing(return_X_y=True, as_frame=True)\n",
    "X = X.drop(columns=[\"Latitude\", \"Longitude\"])\n",
    "\n",
    "# display top 5 rows\n",
    "X.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "74a2f5cc",
   "metadata": {},
   "source": [
    "### Split data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "52282705",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((14448, 6), (6192, 6))"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# let's separate into training and testing set\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    X, y, test_size=0.3, random_state=0)\n",
    "\n",
    "X_train.shape, X_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "adfae3e9",
   "metadata": {},
   "source": [
    "## XGBoost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "89b7867b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-1 {color: black;}#sk-container-id-1 pre{padding: 0;}#sk-container-id-1 div.sk-toggleable {background-color: white;}#sk-container-id-1 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-1 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-1 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-1 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-1 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-1 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-1 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-1 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-1 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-1 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-1 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-1 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-1 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-1 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-1 div.sk-item {position: relative;z-index: 1;}#sk-container-id-1 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-1 div.sk-item::before, #sk-container-id-1 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-1 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-1 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-1 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-1 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-1 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-1 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-1 div.sk-label-container {text-align: center;}#sk-container-id-1 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-1 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-1\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>XGBRegressor(base_score=None, booster=None, callbacks=None,\n",
       "             colsample_bylevel=None, colsample_bynode=None,\n",
       "             colsample_bytree=None, device=None, early_stopping_rounds=None,\n",
       "             enable_categorical=False, eval_metric=None, feature_types=None,\n",
       "             gamma=None, grow_policy=None, importance_type=&#x27;gain&#x27;,\n",
       "             interaction_constraints=None, learning_rate=None, max_bin=None,\n",
       "             max_cat_threshold=None, max_cat_to_onehot=None,\n",
       "             max_delta_step=None, max_depth=2, max_leaves=None,\n",
       "             min_child_weight=None, missing=nan, monotone_constraints=None,\n",
       "             multi_strategy=None, n_estimators=5, n_jobs=None,\n",
       "             num_parallel_tree=None, random_state=3, ...)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-1\" type=\"checkbox\" checked><label for=\"sk-estimator-id-1\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">XGBRegressor</label><div class=\"sk-toggleable__content\"><pre>XGBRegressor(base_score=None, booster=None, callbacks=None,\n",
       "             colsample_bylevel=None, colsample_bynode=None,\n",
       "             colsample_bytree=None, device=None, early_stopping_rounds=None,\n",
       "             enable_categorical=False, eval_metric=None, feature_types=None,\n",
       "             gamma=None, grow_policy=None, importance_type=&#x27;gain&#x27;,\n",
       "             interaction_constraints=None, learning_rate=None, max_bin=None,\n",
       "             max_cat_threshold=None, max_cat_to_onehot=None,\n",
       "             max_delta_step=None, max_depth=2, max_leaves=None,\n",
       "             min_child_weight=None, missing=nan, monotone_constraints=None,\n",
       "             multi_strategy=None, n_estimators=5, n_jobs=None,\n",
       "             num_parallel_tree=None, random_state=3, ...)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "XGBRegressor(base_score=None, booster=None, callbacks=None,\n",
       "             colsample_bylevel=None, colsample_bynode=None,\n",
       "             colsample_bytree=None, device=None, early_stopping_rounds=None,\n",
       "             enable_categorical=False, eval_metric=None, feature_types=None,\n",
       "             gamma=None, grow_policy=None, importance_type='gain',\n",
       "             interaction_constraints=None, learning_rate=None, max_bin=None,\n",
       "             max_cat_threshold=None, max_cat_to_onehot=None,\n",
       "             max_delta_step=None, max_depth=2, max_leaves=None,\n",
       "             min_child_weight=None, missing=nan, monotone_constraints=None,\n",
       "             multi_strategy=None, n_estimators=5, n_jobs=None,\n",
       "             num_parallel_tree=None, random_state=3, ...)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# fit model\n",
    "\n",
    "gbm = XGBRegressor(\n",
    "    importance_type=\"gain\",\n",
    "    n_estimators=5,\n",
    "    max_depth=2,\n",
    "    random_state=3,\n",
    ")\n",
    "\n",
    "gbm.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e8fa192f",
   "metadata": {},
   "source": [
    "## Peek at the ensemble"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "b59c5837",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "booster[0]:\n",
      "0:[MedInc<5] yes=1,no=2,missing=2,gain=6050.60205,cover=14448\n",
      "\t1:[MedInc<3.07990003] yes=3,no=4,missing=4,gain=1505.21838,cover=11310\n",
      "\t\t3:leaf=-0.214339092,cover=5520\n",
      "\t\t4:leaf=0.00459926156,cover=5790\n",
      "\t2:[MedInc<7.80049992] yes=5,no=6,missing=6,gain=1120.00781,cover=3138\n",
      "\t\t5:leaf=0.286677152,cover=2599\n",
      "\t\t6:leaf=0.761799335,cover=539\n",
      "booster[1]:\n",
      "0:[MedInc<5.60379982] yes=1,no=2,missing=2,gain=3219.354,cover=14448\n",
      "\t1:[MedInc<3.55110002] yes=3,no=4,missing=4,gain=1047.51404,cover=12391\n",
      "\t\t3:leaf=-0.130962819,cover=7264\n",
      "\t\t4:leaf=0.0461317673,cover=5127\n",
      "\t2:[MedInc<6.80749989] yes=5,no=6,missing=6,gain=305.747559,cover=2057\n",
      "\t\t5:leaf=0.241858423,cover=1125\n",
      "\t\t6:leaf=0.474598616,cover=932\n",
      "booster[2]:\n",
      "0:[MedInc<4.58330011] yes=1,no=2,missing=2,gain=1641.849,cover=14448\n",
      "\t1:[AveOccup<2.29999995] yes=3,no=4,missing=4,gain=888.338501,cover=10466\n",
      "\t\t3:leaf=0.108552471,cover=2169\n",
      "\t\t4:leaf=-0.107041381,cover=8297\n",
      "\t2:[AveOccup<2.79061365] yes=5,no=6,missing=6,gain=463.413208,cover=3982\n",
      "\t\t5:leaf=0.275801092,cover=1814\n",
      "\t\t6:leaf=0.0702892169,cover=2168\n",
      "booster[3]:\n",
      "0:[MedInc<6.07119989] yes=1,no=2,missing=2,gain=1075.18054,cover=14448\n",
      "\t1:[AveOccup<2.45283008] yes=3,no=4,missing=4,gain=657.481018,cover=12898\n",
      "\t\t3:leaf=0.0810568929,cover=3573\n",
      "\t\t4:leaf=-0.0702768266,cover=9325\n",
      "\t2:[HouseAge<28] yes=5,no=6,missing=6,gain=158.283203,cover=1550\n",
      "\t\t5:leaf=0.156834513,cover=924\n",
      "\t\t6:leaf=0.352460593,cover=626\n",
      "booster[4]:\n",
      "0:[MedInc<2.66179991] yes=1,no=2,missing=2,gain=806.322571,cover=14448\n",
      "\t1:[AveRooms<3.87692308] yes=3,no=4,missing=4,gain=235.82959,cover=3999\n",
      "\t\t3:leaf=0.000298453844,cover=1148\n",
      "\t\t4:leaf=-0.160730481,cover=2851\n",
      "\t2:[AveOccup<2.04825735] yes=5,no=6,missing=6,gain=482.446899,cover=10449\n",
      "\t\t5:leaf=0.254233211,cover=896\n",
      "\t\t6:leaf=0.0241177995,cover=9553\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# print structure of tree\n",
    "\n",
    "booster = gbm.get_booster()\n",
    "booster.dump_model('dump.raw.txt', with_stats = True)\n",
    "\n",
    "with open('dump.raw.txt') as file:\n",
    "    print(file.read())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9efc5a4a",
   "metadata": {},
   "source": [
    "## Feature contribution\n",
    "\n",
    "xgboost does not provide the values at the internal nodes, just at the leaves, so we would need to calculate them manually :/\n",
    "\n",
    "We are not going to do that, we are going to use Eli5 instead, who does the job for us."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9f35f8d4",
   "metadata": {},
   "source": [
    "## Eli5\n",
    "\n",
    "If you have problems to import eli5 with the latest version of sklearn, you need to go to the eli5 files that are stored in your local computer, and change a few lines of code as shown in [this PR](https://github.com/eli5-org/eli5/commit/840695d869e47b8e6cc05baca428d24881113fb6)\n",
    "\n",
    "More details also available [here](https://github.com/eli5-org/eli5/issues/39)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "fbf79991",
   "metadata": {},
   "outputs": [],
   "source": [
    "import eli5"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c1d98379",
   "metadata": {},
   "source": [
    "## Global explanations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "cd662ece",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <style>\n",
       "    table.eli5-weights tr:hover {\n",
       "        filter: brightness(85%);\n",
       "    }\n",
       "</style>\n",
       "\n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "        <table class=\"eli5-weights eli5-feature-importances\" style=\"border-collapse: collapse; border: none; margin-top: 0em; table-layout: auto;\">\n",
       "    <thead>\n",
       "    <tr style=\"border: none;\">\n",
       "        <th style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">Weight</th>\n",
       "        <th style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">Feature</th>\n",
       "    </tr>\n",
       "    </thead>\n",
       "    <tbody>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 80.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.6469\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                MedInc\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 90.71%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.2162\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                AveOccup\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 95.29%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0819\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                AveRooms\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(120, 100.00%, 96.44%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0.0549\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                HouseAge\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                Population\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "        <tr style=\"background-color: hsl(0, 100.00%, 100.00%); border: none;\">\n",
       "            <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "                0\n",
       "                \n",
       "            </td>\n",
       "            <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "                AveBedrms\n",
       "            </td>\n",
       "        </tr>\n",
       "    \n",
       "    \n",
       "    </tbody>\n",
       "</table>\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Feature importance (global)\n",
    "\n",
    "eli5.show_weights(gbm, feature_names=gbm.feature_names_in_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "8804aa36",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MedInc        0.646933\n",
       "AveOccup      0.216249\n",
       "AveRooms      0.081869\n",
       "HouseAge      0.054949\n",
       "AveBedrms     0.000000\n",
       "Population    0.000000\n",
       "dtype: float32"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# let's compare eli's result with that of \n",
    "# xgboost\n",
    "\n",
    "pd.Series(gbm.feature_importances_,\n",
    "          index=gbm.feature_names_in_).sort_values(ascending=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "48ca7404",
   "metadata": {},
   "source": [
    "## Local explanations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "a61fe962",
   "metadata": {},
   "outputs": [
    {
     "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>MedInc</th>\n",
       "      <th>HouseAge</th>\n",
       "      <th>AveRooms</th>\n",
       "      <th>AveBedrms</th>\n",
       "      <th>Population</th>\n",
       "      <th>AveOccup</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>14740</th>\n",
       "      <td>4.1518</td>\n",
       "      <td>22.0</td>\n",
       "      <td>5.663073</td>\n",
       "      <td>1.075472</td>\n",
       "      <td>1551.0</td>\n",
       "      <td>4.180593</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10101</th>\n",
       "      <td>5.7796</td>\n",
       "      <td>32.0</td>\n",
       "      <td>6.107226</td>\n",
       "      <td>0.927739</td>\n",
       "      <td>1296.0</td>\n",
       "      <td>3.020979</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20566</th>\n",
       "      <td>4.3487</td>\n",
       "      <td>29.0</td>\n",
       "      <td>5.930712</td>\n",
       "      <td>1.026217</td>\n",
       "      <td>1554.0</td>\n",
       "      <td>2.910112</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2670</th>\n",
       "      <td>2.4511</td>\n",
       "      <td>37.0</td>\n",
       "      <td>4.992958</td>\n",
       "      <td>1.316901</td>\n",
       "      <td>390.0</td>\n",
       "      <td>2.746479</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15709</th>\n",
       "      <td>5.0049</td>\n",
       "      <td>25.0</td>\n",
       "      <td>4.319261</td>\n",
       "      <td>1.039578</td>\n",
       "      <td>649.0</td>\n",
       "      <td>1.712401</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       MedInc  HouseAge  AveRooms  AveBedrms  Population  AveOccup\n",
       "14740  4.1518      22.0  5.663073   1.075472      1551.0  4.180593\n",
       "10101  5.7796      32.0  6.107226   0.927739      1296.0  3.020979\n",
       "20566  4.3487      29.0  5.930712   1.026217      1554.0  2.910112\n",
       "2670   2.4511      37.0  4.992958   1.316901       390.0  2.746479\n",
       "15709  5.0049      25.0  4.319261   1.039578       649.0  1.712401"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Display a few observations from test set\n",
    "\n",
    "X_test.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "b99a1739",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MedInc           4.151800\n",
       "HouseAge        22.000000\n",
       "AveRooms         5.663073\n",
       "AveBedrms        1.075472\n",
       "Population    1551.000000\n",
       "AveOccup         4.180593\n",
       "Name: 14740, dtype: float64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# pick one observation\n",
    "\n",
    "sample_id = 14740\n",
    "\n",
    "X_test.loc[sample_id]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "ec3ba4d1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <style>\n",
       "    table.eli5-weights tr:hover {\n",
       "        filter: brightness(85%);\n",
       "    }\n",
       "</style>\n",
       "\n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "        \n",
       "\n",
       "    \n",
       "\n",
       "        \n",
       "            \n",
       "                \n",
       "                \n",
       "    \n",
       "        <p style=\"margin-bottom: 0.5em; margin-top: 0em\">\n",
       "            <b>\n",
       "    \n",
       "        y\n",
       "    \n",
       "</b>\n",
       "\n",
       "    \n",
       "    (score <b>-0.102</b>)\n",
       "\n",
       "top features\n",
       "        </p>\n",
       "    \n",
       "    <table class=\"eli5-weights\"\n",
       "           style=\"border-collapse: collapse; border: none; margin-top: 0em; table-layout: auto; margin-bottom: 2em;\">\n",
       "        <thead>\n",
       "        <tr style=\"border: none;\">\n",
       "            \n",
       "                <th style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\" title=\"Feature contribution already accounts for the feature value (for linear models, contribution = weight * feature value), and the sum of feature contributions is equal to the score or, for some classifiers, to the probability. Feature values are shown if &quot;show_feature_values&quot; is True.\">\n",
       "                    Contribution<sup>?</sup>\n",
       "                </th>\n",
       "            \n",
       "            <th style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">Feature</th>\n",
       "            \n",
       "        </tr>\n",
       "        </thead>\n",
       "        <tbody>\n",
       "        \n",
       "            <tr style=\"background-color: hsl(120, 100.00%, 98.01%); border: none;\">\n",
       "    <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "        +0.004\n",
       "    </td>\n",
       "    <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "        MedInc\n",
       "    </td>\n",
       "    \n",
       "</tr>\n",
       "        \n",
       "        \n",
       "\n",
       "        \n",
       "        \n",
       "            <tr style=\"background-color: hsl(0, 100.00%, 99.88%); border: none;\">\n",
       "    <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "        -0.000\n",
       "    </td>\n",
       "    <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "        &lt;BIAS&gt;\n",
       "    </td>\n",
       "    \n",
       "</tr>\n",
       "        \n",
       "            <tr style=\"background-color: hsl(0, 100.00%, 80.00%); border: none;\">\n",
       "    <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "        -0.106\n",
       "    </td>\n",
       "    <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "        AveOccup\n",
       "    </td>\n",
       "    \n",
       "</tr>\n",
       "        \n",
       "\n",
       "        </tbody>\n",
       "    </table>\n",
       "\n",
       "            \n",
       "        \n",
       "\n",
       "        \n",
       "\n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# feature contribution for 1 sample\n",
    "\n",
    "eli5.show_prediction(\n",
    "    gbm, \n",
    "    X_test.loc[sample_id], \n",
    "    feature_names=gbm.feature_names_in_,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "88ce40d2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2.0682462451550387"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train.mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "662024f0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.9662462451550387"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# manual prediction\n",
    "\n",
    "y_train.mean() + 0.004 - 0.106"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "5ef063ce",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1.9657769], dtype=float32)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# prediction of xgboost for that sample\n",
    "\n",
    "gbm.predict(X_test.loc[sample_id].to_frame().T)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "1ab8cbe1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <style>\n",
       "    table.eli5-weights tr:hover {\n",
       "        filter: brightness(85%);\n",
       "    }\n",
       "</style>\n",
       "\n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "        \n",
       "\n",
       "    \n",
       "\n",
       "        \n",
       "            \n",
       "                \n",
       "                \n",
       "    \n",
       "        <p style=\"margin-bottom: 0.5em; margin-top: 0em\">\n",
       "            <b>\n",
       "    \n",
       "        y\n",
       "    \n",
       "</b>\n",
       "\n",
       "    \n",
       "    (score <b>0.553</b>)\n",
       "\n",
       "top features\n",
       "        </p>\n",
       "    \n",
       "    <table class=\"eli5-weights\"\n",
       "           style=\"border-collapse: collapse; border: none; margin-top: 0em; table-layout: auto; margin-bottom: 2em;\">\n",
       "        <thead>\n",
       "        <tr style=\"border: none;\">\n",
       "            \n",
       "                <th style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\" title=\"Feature contribution already accounts for the feature value (for linear models, contribution = weight * feature value), and the sum of feature contributions is equal to the score or, for some classifiers, to the probability. Feature values are shown if &quot;show_feature_values&quot; is True.\">\n",
       "                    Contribution<sup>?</sup>\n",
       "                </th>\n",
       "            \n",
       "            <th style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">Feature</th>\n",
       "            \n",
       "        </tr>\n",
       "        </thead>\n",
       "        <tbody>\n",
       "        \n",
       "            <tr style=\"background-color: hsl(120, 100.00%, 80.00%); border: none;\">\n",
       "    <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "        +0.708\n",
       "    </td>\n",
       "    <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "        MedInc\n",
       "    </td>\n",
       "    \n",
       "</tr>\n",
       "        \n",
       "        \n",
       "\n",
       "        \n",
       "        \n",
       "            <tr style=\"background-color: hsl(0, 100.00%, 99.97%); border: none;\">\n",
       "    <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "        -0.000\n",
       "    </td>\n",
       "    <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "        &lt;BIAS&gt;\n",
       "    </td>\n",
       "    \n",
       "</tr>\n",
       "        \n",
       "            <tr style=\"background-color: hsl(0, 100.00%, 93.09%); border: none;\">\n",
       "    <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "        -0.155\n",
       "    </td>\n",
       "    <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "        AveOccup\n",
       "    </td>\n",
       "    \n",
       "</tr>\n",
       "        \n",
       "\n",
       "        </tbody>\n",
       "    </table>\n",
       "\n",
       "            \n",
       "        \n",
       "\n",
       "        \n",
       "\n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# feature contribution for another sample\n",
    "\n",
    "eli5.show_prediction(\n",
    "    gbm, \n",
    "    X_test.loc[10101], \n",
    "    feature_names=gbm.feature_names_in_,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "e855fd2e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2.621246245155039"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# manual prediction\n",
    "\n",
    "y_train.mean() + 0.708 - 0.155"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "725df449",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([2.6209118], dtype=float32)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# prediction of xgboost for that sample\n",
    "\n",
    "gbm.predict(X_test.loc[10101].to_frame().T)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "15e91778",
   "metadata": {},
   "outputs": [
    {
     "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>target</th>\n",
       "      <th>feature</th>\n",
       "      <th>weight</th>\n",
       "      <th>value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>y</td>\n",
       "      <td>MedInc</td>\n",
       "      <td>0.708015</td>\n",
       "      <td>5.779600</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>y</td>\n",
       "      <td>&lt;BIAS&gt;</td>\n",
       "      <td>-0.000074</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>y</td>\n",
       "      <td>AveOccup</td>\n",
       "      <td>-0.155276</td>\n",
       "      <td>3.020979</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  target   feature    weight     value\n",
       "0      y    MedInc  0.708015  5.779600\n",
       "1      y    <BIAS> -0.000074  1.000000\n",
       "2      y  AveOccup -0.155276  3.020979"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# if we want to obtain a dataframe instead of the \n",
    "# html output\n",
    "\n",
    "eli5.explain_prediction_df(\n",
    "    gbm, \n",
    "    X_test.loc[10101], \n",
    "    feature_names=gbm.feature_names_in_,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f64e0cf5",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "fsml",
   "language": "python",
   "name": "fsml"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.5"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}