examples/supervised/classification-tree.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Classification Tree\n",
    "Implementation (numba-optimized for some parts) of Ordinary Binary Classification Trees (OBCTs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-04-12T17:34:11.263508Z",
     "start_time": "2020-04-12T17:34:11.248507Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>.container {width: 98%}</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%%html\n",
    "<style>.container {width: 98%}</style>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-04-12T17:53:23.501565Z",
     "start_time": "2020-04-12T17:53:21.186516Z"
    }
   },
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-04-12T17:34:25.443706Z",
     "start_time": "2020-04-12T17:34:13.388959Z"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn import datasets\n",
    "from sklearn.metrics import accuracy_score, confusion_matrix\n",
    "from sklearn.model_selection import cross_val_score\n",
    "\n",
    "import sys\n",
    "sys.path.insert(0, \"../\")\n",
    "from fromscratch.supervised.classification_tree import ClassificationTree"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Use a simple dataset (Iris)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-04-12T17:34:58.362343Z",
     "start_time": "2020-04-12T17:34:58.318302Z"
    }
   },
   "outputs": [],
   "source": [
    "data = datasets.load_iris()\n",
    "X = data[\"data\"]\n",
    "y = data[\"target\"].astype(np.float64)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (Over)fit the tree"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-04-12T17:34:59.191574Z",
     "start_time": "2020-04-12T17:34:59.166575Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'expr': 'feature_2 <= 2.45',\n",
       " 'no': {'expr': 'feature_3 <= 1.75',\n",
       "        'no': {'expr': 'feature_2 <= 4.85',\n",
       "               'no': {'label': 2.0},\n",
       "               'yes': {'label': 2.0}},\n",
       "        'yes': {'expr': 'feature_2 <= 4.95',\n",
       "                'no': {'label': 2.0},\n",
       "                'yes': {'label': 1.0}}},\n",
       " 'yes': {'expr': 'feature_0 <= 4.35',\n",
       "         'no': {'expr': 'feature_0 <= 4.45',\n",
       "                'no': {'label': 0.0},\n",
       "                'yes': {'label': 0.0}},\n",
       "         'yes': {'label': 0.0}}}"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tree = ClassificationTree(max_depth=4, min_leaf_samples=1, min_delta_impurity=0)\n",
    "tree.fit(X, y, verbose_fit=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Confusion Matric"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-04-12T17:35:00.038315Z",
     "start_time": "2020-04-12T17:35:00.026311Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[50,  0,  0],\n",
       "       [ 0, 47,  3],\n",
       "       [ 0,  1, 49]], dtype=int64)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred = tree.predict(X)\n",
    "confusion_matrix(y, y_pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Cross-validated accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-04-12T17:35:00.687598Z",
     "start_time": "2020-04-12T17:35:00.605579Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.96"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(cross_val_score(tree, X, y, cv=5))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-25T20:06:02.803118Z",
     "start_time": "2020-03-25T20:06:02.775123Z"
    }
   },
   "source": [
    "### Plot 2 dimensions of the data and the corresponding decision boundary of the tree\n",
    "We can see the rectangular decision regions created by the tree"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-04-12T17:35:07.334311Z",
     "start_time": "2020-04-12T17:35:06.788308Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "f1, f2 = 2, 3\n",
    "\n",
    "plt.figure(figsize=(7, 7))\n",
    "\n",
    "cm = plt.cm.Spectral\n",
    "\n",
    "h = .05\n",
    "x_min, x_max = X[:, f1].min() - .5, X[:, f1].max() + .5\n",
    "y_min, y_max = X[:, f2].min() - .5, X[:, f2].max() + .5\n",
    "xx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max, h))\n",
    "\n",
    "tree = ClassificationTree(max_depth=5, min_leaf_samples=1, min_delta_impurity=0.0)\n",
    "tree.fit(X[:, [f1, f2]], y)\n",
    "\n",
    "Z = tree.predict(np.c_[xx.ravel(), yy.ravel()])\n",
    "Z = Z.reshape(xx.shape)\n",
    "plt.contourf(xx, yy, Z, cmap=cm, alpha=.8)\n",
    "\n",
    "plt.scatter(X[:, f1], X[:, f2], c=y, cmap=cm, edgecolor=\"grey\")\n",
    "\n",
    "plt.xlabel(f\"feature {f1+1}\")\n",
    "plt.ylabel(f\"feature {f2+1}\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-25T23:25:22.519801Z",
     "start_time": "2020-03-25T23:25:22.475939Z"
    }
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.7.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}