notebooks/ensemble_sol_05.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "972cffb2",
   "metadata": {},
   "source": [
    "# 📃 Solution for Exercise M6.05\n",
    "\n",
    "The aim of the exercise is to get familiar with the histogram\n",
    "gradient-boosting in scikit-learn. Besides, we will use this model within\n",
    "a cross-validation framework in order to inspect internal parameters found\n",
    "via grid-search.\n",
    "\n",
    "We will use the California housing dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "2f0d4d1a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-21T12:17:47.997089Z",
     "start_time": "2021-06-21T12:17:47.575166Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.datasets import fetch_california_housing\n",
    "\n",
    "data, target = fetch_california_housing(return_X_y=True, as_frame=True)\n",
    "target *= 100  # rescale the target in k$"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "15663a12",
   "metadata": {},
   "source": [
    "First, create a histogram gradient boosting regressor. You can set the\n",
    "trees number to be large, and configure the model to use early-stopping."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "3eba2b76",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-21T12:17:50.135660Z",
     "start_time": "2021-06-21T12:17:50.093388Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.experimental import enable_hist_gradient_boosting\n",
    "from sklearn.ensemble import HistGradientBoostingRegressor\n",
    "\n",
    "hist_gbdt = HistGradientBoostingRegressor(\n",
    "    max_iter=1000, early_stopping=True, random_state=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4755ff1c",
   "metadata": {},
   "source": [
    "We will use a grid-search to find some optimal parameter for this model.\n",
    "In this grid-search, you should search for the following parameters:\n",
    "\n",
    "* `max_depth: [3, 8]`;\n",
    "* `max_leaf_nodes: [15, 31]`;\n",
    "* `learning_rate: [0.1, 1]`.\n",
    "\n",
    "Feel free to explore the space with additional values. Create the\n",
    "grid-search providing the previous gradient boosting instance as the model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "8a53e3c8",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-21T12:17:50.818672Z",
     "start_time": "2021-06-21T12:17:50.816444Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "params = {\n",
    "    \"max_depth\": [3, 8],\n",
    "    \"max_leaf_nodes\": [15, 31],\n",
    "    \"learning_rate\": [0.1, 1],\n",
    "}\n",
    "\n",
    "search = GridSearchCV(hist_gbdt, params)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c579bde7",
   "metadata": {},
   "source": [
    "Finally, we will run our experiment through cross-validation. In this regard,\n",
    "define a 5-fold cross-validation. Besides, be sure to shuffle the data.\n",
    "Subsequently, use the function `sklearn.model_selection.cross_validate`\n",
    "to run the cross-validation. You should also set `return_estimator=True`,\n",
    "so that we can investigate the inner model trained via cross-validation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "32ebb5ed",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-21T12:18:09.832174Z",
     "start_time": "2021-06-21T12:17:53.096294Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import cross_validate\n",
    "from sklearn.model_selection import KFold\n",
    "\n",
    "cv = KFold(n_splits=5, shuffle=True, random_state=0)\n",
    "results = cross_validate(\n",
    "    search, data, target, cv=cv, return_estimator=True, n_jobs=-1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6c3c1777",
   "metadata": {},
   "source": [
    "Now that we got the cross-validation results, print out the mean and\n",
    "standard deviation score."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "5a64803e",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-21T12:18:09.842570Z",
     "start_time": "2021-06-21T12:18:09.837193Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "R2 score with cross-validation:\n",
      "0.839 +/- 0.006\n"
     ]
    }
   ],
   "source": [
    "print(f\"R2 score with cross-validation:\\n\"\n",
    "      f\"{results['test_score'].mean():.3f} +/- \"\n",
    "      f\"{results['test_score'].std():.3f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4b556093",
   "metadata": {},
   "source": [
    "Then inspect the `estimator` entry of the results and check the best\n",
    "parameters values. Besides, check the number of trees used by the model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "4338dce5",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-21T12:18:11.267378Z",
     "start_time": "2021-06-21T12:18:11.264236Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'learning_rate': 0.1, 'max_depth': 3, 'max_leaf_nodes': 15}\n",
      "# trees: 528\n",
      "{'learning_rate': 0.1, 'max_depth': 8, 'max_leaf_nodes': 15}\n",
      "# trees: 447\n",
      "{'learning_rate': 0.1, 'max_depth': 3, 'max_leaf_nodes': 15}\n",
      "# trees: 576\n",
      "{'learning_rate': 0.1, 'max_depth': 8, 'max_leaf_nodes': 15}\n",
      "# trees: 290\n",
      "{'learning_rate': 0.1, 'max_depth': 8, 'max_leaf_nodes': 15}\n",
      "# trees: 414\n"
     ]
    }
   ],
   "source": [
    "for estimator in results[\"estimator\"]:\n",
    "    print(estimator.best_params_)\n",
    "    print(f\"# trees: {estimator.best_estimator_.n_iter_}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c500fb69",
   "metadata": {},
   "source": [
    "We observe that the parameters are varying. We can get the intuition that\n",
    "results of the inner CV are very close for certain set of parameters."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "46ccdae0",
   "metadata": {},
   "source": [
    "Inspect the results of the inner CV for each estimator of the outer CV.\n",
    "Aggregate the mean test score for each parameter combination and make a box\n",
    "plot of these scores."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "deaf8fa7",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-21T12:20:07.087977Z",
     "start_time": "2021-06-21T12:20:07.056491Z"
    }
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "index_columns = [f\"param_{name}\" for name in params.keys()]\n",
    "columns = index_columns + [\"mean_test_score\"]\n",
    "\n",
    "inner_cv_results = []\n",
    "for cv_idx, estimator in enumerate(results[\"estimator\"]):\n",
    "    search_cv_results = pd.DataFrame(estimator.cv_results_)\n",
    "    search_cv_results = search_cv_results[columns].set_index(index_columns)\n",
    "    search_cv_results = search_cv_results.rename(\n",
    "        columns={\"mean_test_score\": f\"CV {cv_idx}\"})\n",
    "    inner_cv_results.append(search_cv_results)\n",
    "inner_cv_results = pd.concat(inner_cv_results, axis=1).T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "10d39255",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-21T12:20:08.448552Z",
     "start_time": "2021-06-21T12:20:08.167394Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "color = {\"whiskers\": \"black\", \"medians\": \"black\", \"caps\": \"black\"}\n",
    "inner_cv_results.plot.box(vert=False, color=color)\n",
    "plt.xlabel(\"R2 score\")\n",
    "plt.ylabel(\"Parameters\")\n",
    "_ = plt.title(\"Inner CV results with parameters\\n\"\n",
    "              \"(max_depth, max_leaf_nodes, learning_rate)\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "686829af",
   "metadata": {},
   "source": [
    "We see that the first 4 ranked set of parameters are very close.\n",
    "We could select any of these 4 combinations.\n",
    "It coincides with the results we observe when inspecting the\n",
    "best parameters of the outer CV."
   ]
  }
 ],
 "metadata": {
  "jupytext": {
   "cell_metadata_filter": "-all",
   "main_language": "python",
   "notebook_metadata_filter": "-all"
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.4"
  },
  "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": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}