Lecture17_Ensemble_1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Ensemble"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- &#51068;&#47144;&#51032; &#50696;&#52769;&#44592; (&#51593;, &#48516;&#47448;&#45208; &#54924;&#44480; &#47784;&#45944;)&#47196;&#48512;&#53552; &#50696;&#52769;&#51012; &#49688;&#51665;&#54616;&#47732; &#44032;&#51109; &#51339;&#51008; &#47784;&#45944; &#54616;&#45208;&#48372;&#45796; &#45908; &#51339;&#51008; &#50696;&#52769;&#51012; &#50619;&#51012; &#49688; &#51080;&#45796;. &#51068;&#47144;&#51032; &#50696;&#52769;&#44592;&#47484; **ensemble**&#51060;&#46972;&#44256; &#48512;&#47476;&#44592; &#46412;&#47928;&#50640; &#51060;&#47484; **ensemble learning**&#51060;&#46972;&#44256; &#54620;&#45796;. &#50521;&#49345;&#48660; &#54617;&#49845; &#50508;&#44256;&#47532;&#51608;&#51012; **ensemble method**&#46972;&#44256; &#54620;&#45796;.\n",
    "- &#50521;&#49345;&#48660; &#54617;&#49845; &#50508;&#44256;&#47532;&#51608;&#51032; &#50696;&#49884; \n",
    "    - &#54984;&#47144; &#49464;&#53944;&#47196;&#48512;&#53552; &#47924;&#51089;&#50948;&#47196; &#44033;&#44592; &#45796;&#47480; &#49436;&#48652;&#49483;&#51012; &#49373;&#49457;&#54620;&#45796;.\n",
    "    - &#44033;&#44033;&#51032; &#49436;&#48652;&#49483;&#50640; &#51068;&#47144;&#51032; &#44208;&#51221; &#53944;&#47532; &#48516;&#47448;&#44592;&#47484; &#54984;&#47144;&#49884;&#53416;&#45796;. \n",
    "    - &#47784;&#46304; &#44060;&#48324; &#53944;&#47532;&#51032; &#50696;&#52769;&#51012; &#44396;&#54620;&#45796;. \n",
    "    - &#44536;&#47088; &#45796;&#51020; &#44032;&#51109; &#47566;&#51008; &#49440;&#53469;(major vote)&#51012; &#48155;&#51008; &#53364;&#47000;&#49828;&#47484; &#50696;&#52769; &#44208;&#44284;&#44050;&#51004;&#47196; &#54620;&#45796;. \n",
    "- &#44208;&#51221; &#53944;&#47532;&#51032; &#45824;&#54364;&#51201;&#51064; ensemble&#50640;&#45716; **Random Forest**&#44032; &#51080;&#45796;. &#44036;&#45800;&#54620; &#48169;&#48277;&#51060;&#51648;&#47564; Random Forest&#45716; &#50724;&#45720;&#45216; &#44032;&#51109; &#44053;&#47141;&#54620; &#47672;&#49888;&#47084;&#45789; &#50508;&#44256;&#47532;&#51608; &#51473; &#54616;&#45208;&#51060;&#47728;, &#54788;&#50629;&#50640;&#49436;&#46020; &#47566;&#51060; &#49324;&#50857;&#46108;&#45796;.\n",
    "- &#45824;&#54364;&#51201;&#51064; ensemble &#44592;&#48277;\n",
    " - Bagging\n",
    " - Boosting\n",
    " - Stacking(ensemble&#51032; ensemble)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "# Ensemble &#44592;&#48277;&#51012; &#54876;&#50857;&#54620; &#48516;&#47448;&#44592; (&#51060;&#48264; &#44053;&#51032; &#45432;&#53944;&#50640;&#49436; &#48176;&#50864;&#44172; &#46104;&#45716;)\n",
    "### 1. Voting based classifer\n",
    "### 2. Bagging\n",
    "### 3. Random Forest\n",
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1. Voting based classifier (&#53804;&#54364; &#44592;&#48152; &#48516;&#47448;&#44592;)\n",
    " - &#45908; &#51339;&#51008; &#48516;&#47448;&#44592;&#47484; &#47564;&#46300;&#45716; &#48169;&#48277;&#51008; &#44033; &#48516;&#47448;&#44592;&#51032; &#50696;&#52769;&#51012; &#47784;&#50500;&#49436; &#44032;&#51109; &#47566;&#51060; &#49440;&#53469;&#46108; &#53364;&#47000;&#49828;&#47484; &#50696;&#52769;&#54616;&#45716; &#44163;\n",
    " - **Hard voting** or **Soft voting** classifier : &#45796;&#49688;&#44208; &#53804;&#54364;&#47196; &#51221;&#54644;&#51648;&#45716; &#48516;&#47448;&#44592;\n",
    "  - &#44033; &#44060;&#48324; &#48516;&#47448;&#44592;(base classifier) &#51473; &#44032;&#51109; &#46832;&#50612;&#45212; &#48516;&#47448;&#44592; &#48372;&#45796;&#46020;, &#45796;&#49688;&#44208; &#53804;&#54364; &#48516;&#47448;&#44592;&#51032; &#51221;&#54869;&#46020;&#44032; &#45458;&#51008; &#44221;&#50864;&#44032; &#51316;&#51116;\n",
    "  - **Base classifer&#44032; week learner(&#50557;&#54620; &#54617;&#49845;&#44592;, &#47004;&#45924; &#52628;&#52769;&#48372;&#45796; &#51312;&#44552; &#45908; &#51339;&#51008; &#49457;&#45733;&#51012; &#45236;&#45716; &#48516;&#47448;&#44592;)&#51068;&#51648;&#46972;&#46020;, &#52649;&#48516;&#54616;&#44172; &#47566;&#44256; &#45796;&#50577;&#54616;&#45796;&#47732; strong learner(&#44053;&#54620; &#54617;&#49845;&#44592;, &#45458;&#51008; &#51221;&#54869;&#46020;&#47484; &#45236;&#45716; &#48516;&#47448;&#44592;)&#44032; &#46112; &#49688; &#51080;&#51020;**\n",
    "  - Law of large numbers(&#53360; &#49688;&#51032; &#48277;&#52825;)\n",
    "    - &#46041;&#51204; &#45912;&#51648;&#44592; &#50696;&#51228; : &#50526;&#47732;&#51060; &#45208;&#50732; &#54869;&#47456;&#51060; 51%&#51060;&#44256; &#46263;&#47732;&#51060; &#45208;&#50732; &#54869;&#47456;&#51060; 49%&#51064; &#46041;&#51204;&#51060; &#51316;&#51116;&#54620;&#45796;&#44256; &#54616;&#51088;. &#46041;&#51204;&#51012; 1,000&#48264; &#45912;&#51276;&#51012; &#46412;, &#50526;&#47732;&#51060; &#45796;&#49688;&#44032; &#45208;&#50732; &#54869;&#47456;&#51008; &#50620;&#47560;&#51068;&#44620;? \n",
    "    - &#51060;&#54637;&#48516;&#54252;&#47484; &#49324;&#50857;&#54616;&#51088;. &#49457;&#44277; &#54869;&#47456;&#51060; 51%&#51064; &#46041;&#51204;&#51012; 1,000&#48264; &#45912;&#51276;&#51012; &#46412; &#50526;&#47732;&#51060; 1&#48264;&#47564; &#45208;&#50732; &#54869;&#47456;&#51008; $\\frac{1000}{1} \\times 0.51^{1} \\times(1-0.51)^{1000-1}=1.6\\times 10^{-307}$&#51060;&#45796;. &#51060;&#47088;&#49885;&#51004;&#47196; 499&#44620;&#51648;&#51032; &#54869;&#47456;&#51012; &#45908;&#54616;&#44256; 1&#50640;&#49436; &#48764;&#47732; &#46108;&#45796;. (&#51593; 1&#50640;&#49436; &#46263;&#47732;&#51060; &#45796;&#49688;&#44032; &#45208;&#50728; &#44221;&#50864;&#51032; &#54869;&#47456;&#46308;&#51012; &#47784;&#46160; &#48764;&#51452;&#45716; &#44163;&#51060;&#45796;.)\n",
    "    - 51% &#51221;&#54869;&#46020;&#47484; &#44032;&#51652; 1,000&#44060;&#51032; &#48516;&#47448;&#44592;&#47196; &#50521;&#49345;&#48660; &#47784;&#45944;&#51012; &#44396;&#52629;&#54620;&#45796;&#47732; 75%&#51032; &#51221;&#54869;&#46020;&#47484; &#44592;&#45824;&#54624; &#49688; &#51080;&#45796;. (&#48516;&#47448;&#44592;&#44032; &#50756;&#48317;&#54616;&#44172; &#46021;&#47549; & &#50724;&#52264;&#51032; &#49345;&#44288;&#44288;&#44228;&#44032; &#50630;&#50612;&#50556; &#44032;&#45733;)\n",
    "  - Ensemble &#48169;&#48277;&#51008; &#50696;&#52769;&#44592;&#44032; &#44032;&#45733;&#54620; &#54620; &#49436;&#47196; &#46021;&#47549;&#51201;&#51068; &#46412; &#52572;&#44256;&#51032; &#49457;&#45733;&#51012; &#48156;&#55064;&#54620;&#45796;. &#45796;&#50577;&#54620; &#48516;&#47448;&#44592;&#47484; &#50619;&#45716; &#54620; &#44032;&#51648; &#48169;&#48277;&#51008; **&#44033;&#44592; &#45796;&#47480; &#50508;&#44256;&#47532;&#51608;&#51004;&#47196; &#54617;&#49845;**&#49884;&#53412;&#45716; &#44163;&#51060;&#45796;. &#51060;&#47111;&#44172; &#54616;&#47732; &#47588;&#50864; &#45796;&#47480; &#51333;&#47448;&#51032; &#50724;&#52264;&#47484; &#47564;&#46308; &#44032;&#45733;&#49457;&#51060; &#45458;&#44592; &#46412;&#47928;&#50640; &#50521;&#49345;&#48660; &#47784;&#45944;&#51032; &#51221;&#54869;&#46020;&#47484; &#54693;&#49345;&#49884;&#53420; &#49688; &#51080;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.7467502275561786\n",
      "0.9777976478701533\n"
     ]
    }
   ],
   "source": [
    "# &#46041;&#51204; &#45912;&#51648;&#44592; &#50696;&#51228;\n",
    "from scipy import stats\n",
    "\n",
    "print(1 - stats.binom.cdf(499, 1000, 0.51))  # &#52380; &#48264; &#45912;&#51276;&#51012; &#46412; &#50526;&#47732;&#51060; &#51208;&#48152; &#51060;&#49345; &#45208;&#50732; &#54869;&#47456;\n",
    "print(1 - stats.binom.cdf(4999, 10000, 0.51))  # &#47564; &#48264; &#45912;&#51276;&#51012; &#46412; &#50526;&#47732;&#51060; &#51208;&#48152; &#51060;&#49345; &#45208;&#50732; &#54869;&#47456;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### &#45936;&#51060;&#53552; &#47196;&#46377;\n",
    "Skleran&#50640;&#49436;&#45716; &#53804;&#54364; &#44592;&#48152; &#48516;&#47448;&#44592;&#47484; `VotingClassifier`&#47196; &#51228;&#44277;&#54616;&#44256;, base learner&#47196; &#49436;&#47196; &#45796;&#47480; &#47784;&#45944;&#46308;&#51012; &#49324;&#50857;&#54624; &#49688; &#51080;&#46020;&#47197; &#54644;&#51456;&#45796;.\n",
    "\n",
    "Sklearn&#50640;&#49436; &#51228;&#44277;&#54644;&#51452;&#45716; moon &#45936;&#51060;&#53552;&#47484; &#54876;&#50857;&#54644;&#48372;&#51088;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sklearn.datasets as datasets\n",
    "x, y = datasets.make_moons(n_samples=1000, noise=0.3, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.17186341,  0.59624885],\n",
       "       [ 1.25328273, -0.26541353],\n",
       "       [ 0.72322405,  0.2319425 ],\n",
       "       [-0.06519764, -0.65519443],\n",
       "       [-0.79949325,  0.55293543],\n",
       "       [ 0.71131439,  0.07365753],\n",
       "       [ 0.26252452, -0.17853072],\n",
       "       [-0.17500737,  0.5947448 ],\n",
       "       [ 1.54052117, -0.36735281],\n",
       "       [ 0.81833812,  1.06585092]])"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 1, 1, 1, 0, 1, 1, 1, 1, 0])"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y[:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### &#54617;&#49845; / &#53580;&#49828;&#53944;&#49483; &#44396;&#48516;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "train_x, test_x, train_y, test_y = train_test_split(x, y, random_state=42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <font color=blue>&#53804;&#54364; &#44592;&#48152; &#48516;&#47448;&#44592; &#54617;&#49845;</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> **Hard voting** : Base learner&#51032; &#44208;&#44284;&#44050;&#51060; &#53364;&#47000;&#49828;&#51060;&#47728;, major class&#47484; &#49324;&#50857;&#54616;&#50668; &#52572;&#51333; &#44208;&#44284;&#47484; &#46020;&#52636;&#54620;&#45796;."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Base learner\n",
    " - LogisticRegression\n",
    " - RandomForestClassifier\n",
    " - SVC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import VotingClassifier\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "# Base learner&#46308;\n",
    "log_clf = LogisticRegression(solver=\"liblinear\", random_state=42)\n",
    "dt_7_clf = DecisionTreeClassifier(max_depth=7, random_state=42)\n",
    "dt_10_clf = DecisionTreeClassifier(max_depth=10, random_state=42)\n",
    "\n",
    "voting_clf = VotingClassifier(\n",
    "    estimators=[('lr', log_clf), ('dt_7', dt_7_clf), ('dt_10', dt_10_clf)], voting='hard'\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "LogisticRegression 0.852\n",
      "DecisionTreeClassifier 0.892\n",
      "DecisionTreeClassifier 0.892\n",
      "VotingClassifier 0.908\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.6/site-packages/sklearn/preprocessing/label.py:151: DeprecationWarning: The truth value of an empty array is ambiguous. Returning False, but in future this will result in an error. Use `array.size > 0` to check that an array is not empty.\n",
      "  if diff:\n"
     ]
    }
   ],
   "source": [
    "# &#44060;&#48324; &#54617;&#49845;&#44592;&#51032; &#49457;&#45733; (&#47004;&#45924;&#54252;&#47112;&#49828;&#53944;, &#47196;&#51648;&#49828;&#54001;, SVM, &#45796;&#49688;&#44208; &#44592;&#48152; &#48516;&#47448;&#44592;)\n",
    "for clf in (log_clf, dt_7_clf, dt_10_clf, voting_clf):\n",
    "    clf.fit(train_x, train_y)\n",
    "    test_pred = clf.predict(test_x)\n",
    "    print(clf.__class__.__name__, accuracy_score(test_y, test_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> **Soft voting** : Base learner&#51032; &#44208;&#44284;&#44050;&#51060; &#54869;&#47456;&#51060;&#47728;, &#44033; learner&#51032; &#54869;&#47456;&#51032; &#54217;&#44512;&#51012; &#45236;&#50612; &#52572;&#51333; &#44208;&#44284;&#47484; &#46020;&#52636;&#54620;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "log_clf = LogisticRegression(solver=\"liblinear\", random_state=42)\n",
    "dt_7_clf = DecisionTreeClassifier(max_depth=7, random_state=42)\n",
    "dt_10_clf = DecisionTreeClassifier(max_depth=10, random_state=42)\n",
    "\n",
    "voting_clf = VotingClassifier(\n",
    "    estimators=[('lr', log_clf), ('dt_7', dt_7_clf), ('dt_10', dt_10_clf)], voting='soft'\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "LogisticRegression 0.852\n",
      "DecisionTreeClassifier 0.892\n",
      "DecisionTreeClassifier 0.892\n",
      "VotingClassifier 0.908\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.6/site-packages/sklearn/preprocessing/label.py:151: DeprecationWarning: The truth value of an empty array is ambiguous. Returning False, but in future this will result in an error. Use `array.size > 0` to check that an array is not empty.\n",
      "  if diff:\n"
     ]
    }
   ],
   "source": [
    "# &#44060;&#48324; &#54617;&#49845;&#44592;&#51032; &#49457;&#45733;\n",
    "for clf in (log_clf, dt_7_clf, dt_10_clf, voting_clf):\n",
    "    clf.fit(train_x, train_y)\n",
    "    test_pred = clf.predict(test_x)\n",
    "    print(clf.__class__.__name__, accuracy_score(test_y, test_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2. Bagging and Pasting"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- **Bagging (Boostrap Aggregating)**\n",
    " - <font color=red>&#54984;&#47144; &#49464;&#53944;&#50640;&#49436; &#51473;&#48373;&#51012; &#54728;&#50857;&#54616;&#50668; &#49368;&#54540;&#47553;&#51012; &#54616;&#44256;, &#44536; &#54984;&#47144; &#49464;&#53944;&#50640;&#49436; 1&#44060;&#51032; &#54617;&#49845; &#50508;&#44256;&#47532;&#51608;&#51012; &#54984;&#47144; &#49884;&#53416; &#44208;&#44284;&#47484; &#51312;&#54633;&#54616;&#45716; &#44592;&#48277;</font>\n",
    " - &#47784;&#46304; &#50696;&#52769;&#44592;&#44032; &#54984;&#47144;&#51012; &#47560;&#52824;&#47732; &#50521;&#49345;&#48660;&#51008; &#47784;&#46304; &#50696;&#52769;&#44592;&#51032; &#50696;&#52769;&#51012; &#47784;&#50500;&#49436; &#49352;&#47196;&#50868; &#49368;&#54540;&#50640; &#45824;&#54620; &#50696;&#52769;&#51012; &#49373;&#49457;\n",
    " - &#48516;&#47448;&#51032; &#44221;&#50864;&#45716; &#53685;&#44228;&#51201; &#52572;&#48712;&#44050;(&#51593;, hard voting classifier &#52376;&#47100; &#44032;&#51109; &#47566;&#51008; &#50696;&#52769; &#44208;&#44284;)&#51012; &#49324;&#50857;\n",
    " - &#54924;&#44480;&#51032; &#44221;&#50864;&#45716; &#54217;&#44512;&#51012; &#49324;&#50857;\n",
    " - <font color=red>&#44060;&#48324; &#50696;&#52769;&#44592;&#45716; &#50896;&#48376; &#54984;&#47144; &#49464;&#53944;&#47196; &#54984;&#47144;&#49884;&#53416; &#44163;&#48372;&#45796; &#55016;&#50476; &#53356;&#44172; &#54200;&#54693;&#46104;&#50612; &#51080;&#51648;&#47564; &#50521;&#49345;&#48660;&#51032; &#44208;&#44284; &#54200;&#54693;&#44284; &#48516;&#49328;&#51060; &#47784;&#46160; &#44048;&#49548; (&#51068;&#48152;&#51201;&#51004;&#47196; &#50521;&#49345;&#48660; &#44592;&#48277;&#50640;&#49436;&#45716; &#54200;&#54693;&#51008; &#48708;&#49847;&#54616;&#51648;&#47564;, &#48516;&#49328;&#51008; &#51460;&#50612;&#46316;)</font>\n",
    " - &#48337;&#47148;&#47196; &#54617;&#49845;&#51060; &#44032;&#45733;\n",
    "- Pasting\n",
    " - &#54984;&#47144; &#49464;&#53944;&#50640;&#49436; &#51473;&#48373;&#51012; &#54728;&#50857;&#54616;&#51648; &#50506;&#44256; &#49368;&#54540;&#47553;&#51012; &#54616;&#44256;, &#44536; &#54984;&#47144; &#49464;&#53944;&#50640;&#49436; 1&#44060;&#51032; &#54617;&#49845; &#50508;&#44256;&#47532;&#51608;&#51012; &#54984;&#47144; &#49884;&#53416; &#44208;&#44284;&#47484; &#51312;&#54633;&#54616;&#45716; &#44592;&#48277;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <font color=blue>Bagging &#54617;&#49845;</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import BaggingClassifier\n",
    "\n",
    "bag_clf = BaggingClassifier(\n",
    "    DecisionTreeClassifier(random_state=42), n_estimators=500,\n",
    "    max_samples=100, bootstrap=True, n_jobs=-1\n",
    ")\n",
    "\n",
    "bag_clf.fit(train_x, train_y)\n",
    "pred_y = bag_clf.predict(test_x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.912"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(test_y, pred_y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### &#45800;&#51068; Decision tree (&#50948;&#51032; &#44208;&#44284;&#50752; &#48708;&#44368;&#50857;)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "tree = DecisionTreeClassifier(random_state=42)\n",
    "tree.fit(train_x, train_y)\n",
    "pred_y_tree = tree.predict(test_x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.892"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(test_y, pred_y_tree)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "*(Advanced)*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "from matplotlib.colors import ListedColormap\n",
    "import numpy as np\n",
    "\n",
    "# Classifer&#47484; &#51077;&#47141;&#51004;&#47196; &#48155;&#44256;, &#44208;&#51221;&#44221;&#44228;&#47484; &#44536;&#47140;&#51452;&#45716; Help &#54632;&#49688; &#51077;&#45768;&#45796;.\n",
    "def plot_decision_boundary(clf, X, y, axes=[-1.5, 2.5, -1, 1.5], alpha=0.5, contour=True):\n",
    "    x1s = np.linspace(axes[0], axes[1], 100)\n",
    "    x2s = np.linspace(axes[2], axes[3], 100)\n",
    "    x1, x2 = np.meshgrid(x1s, x2s)\n",
    "    X_new = np.c_[x1.ravel(), x2.ravel()]\n",
    "    y_pred = clf.predict(X_new).reshape(x1.shape)\n",
    "    custom_cmap = ListedColormap(['#fafab0','#9898ff','#a0faa0'])\n",
    "    plt.contourf(x1, x2, y_pred, alpha=0.3, cmap=custom_cmap)\n",
    "    if contour:\n",
    "        custom_cmap2 = ListedColormap(['#7d7d58','#4c4c7f','#507d50'])\n",
    "        plt.contour(x1, x2, y_pred, cmap=custom_cmap2, alpha=0.8)\n",
    "    plt.plot(X[:, 0][y==0], X[:, 1][y==0], \"yo\", alpha=alpha)\n",
    "    plt.plot(X[:, 0][y==1], X[:, 1][y==1], \"bs\", alpha=alpha)\n",
    "    plt.axis(axes)\n",
    "    plt.xlabel(r\"$x_1$\", fontsize=18)\n",
    "    plt.ylabel(r\"$x_2$\", fontsize=18, rotation=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a1d8b8dd8>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(11, 4))\n",
    "plt.subplot(121)\n",
    "plot_decision_boundary(tree, x, y)\n",
    "plt.title('Decision Tree', fontsize=14)\n",
    "plt.subplot(122)\n",
    "plot_decision_boundary(bag_clf, x, y)\n",
    "plt.title('Decision Trees with Bagging', fontsize=14)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Bagging&#51032; &#44221;&#50864; &#51068;&#48152;&#54868;&#51032; &#49457;&#45733;&#51060; &#45908; &#46832;&#50612;&#45224; (&#44208;&#51221;&#44221;&#44228;&#47484; &#49332;&#54196;&#48372;&#47732;)\n",
    "\n",
    "- **<font color=red>Bootstraping&#51008; &#44033; &#50696;&#52769;&#44592;&#44032; &#54617;&#49845;&#54616;&#45716; &#49436;&#48652;&#49483;&#50640; &#45796;&#50577;&#49457;&#51012; &#51613;&#44032;&#49884;&#53412;&#48064;&#47196; bagging&#51060; pasting&#48372;&#45796; bias&#44032; &#51312;&#44552; &#45908; &#45458;&#51020;. &#54616;&#51648;&#47564; &#51060;&#45716; learner&#46308;&#51032; &#49345;&#44288;&#44288;&#44228;&#47484; &#51460;&#51060;&#48064;&#47196; &#50521;&#49345;&#48660;&#51032; &#48516;&#49328;&#51012; &#44048;&#49548;&#49884;&#53428;</font>**\n",
    "- &#51204;&#48152;&#51201;&#51004;&#47196; bagging&#51012; &#49440;&#54840; (bagging&#44284; pasting&#51012; &#47784;&#46160; &#54217;&#44032;&#54644;&#49436; &#45908; &#45208;&#51008; &#51901;&#51012; &#45817;&#50672;&#55176; &#49440;&#53469;)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### OOB (out-of-bag)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "&#48176;&#44613;&#51012; &#49324;&#50857;&#54616;&#47732; &#50612;&#46500; &#49368;&#54540;&#51008; &#54620; &#50696;&#52769;&#44592;&#47484; &#50948;&#54644; &#50668;&#47084; &#48264; &#49368;&#54540;&#47553;&#46104;&#44256; &#50612;&#46500; &#44163;&#51008; &#51204;&#54784; &#49440;&#53469;&#46104;&#51648; &#50506;&#51012; &#44032;&#45733;&#49457;&#51060; &#51316;&#51116;&#54620;&#45796;. BaggingClasiifer&#45716; &#44592;&#48376;&#44050;&#51004;&#47196; &#51473;&#48373;&#51012; &#54728;&#50857;&#54616;&#50668;(with replacement) &#54984;&#47144; &#49464;&#53356;&#51032; &#53356;&#44592; &#47564;&#53372;&#51064; m&#44060; &#49368;&#54540;&#51012; &#49440;&#53469;&#54620;&#45796;. &#51060;&#45716; <U>&#54217;&#44512;&#51201;&#51004;&#47196; &#44033; &#50696;&#52769;&#44592;&#50640; &#54984;&#47144; &#49368;&#54540;&#51032; 63% &#51221;&#46020;&#47564; &#49368;&#54540;&#47553;&#46108;&#45796;&#45716; &#44163;&#51012; &#51032;&#48120;</U>&#54620;&#45796;.\n",
    " - m&#44060;&#51032; &#49368;&#54540;&#50640;&#49436; &#47924;&#51089;&#50948;&#47196; &#54616;&#45208;&#47484; &#52628;&#52636;&#54624; &#46412; &#49440;&#53469;&#46104;&#51648; &#50506;&#51012; &#54869;&#47456;&#51060; $1-\\frac{1}{m}$&#51060;&#44256; &#51060;&#47484; $m$&#48264; &#48152;&#48373;&#54664;&#51012; &#46412;&#46020; &#49440;&#53469;&#46104;&#51648; &#50506;&#51012; &#54869;&#47456;&#51008; $(1-\\frac{1}{m})^m$. &#47196;&#44536;&#50752; &#47196;&#54588;&#53448; &#51221;&#47532;&#47484; &#51060;&#50857;&#54616;&#47732; $e^{-1}$&#44284; &#44057;&#50500;&#51652;&#45796;. &#46384;&#46972;&#49436; &#49368;&#54540;&#47553;&#46112; &#54869;&#47456;&#51008; $1-e^{-1}$&#51060; &#46108;&#45796;.\n",
    " \n",
    "&#50668;&#44592;&#49436; &#49440;&#53469;&#46104;&#51648; &#50506;&#51008; &#54984;&#47144; &#49368;&#54540;&#51032; &#45208;&#47672;&#51648; 37%&#47484; **oob &#49368;&#54540;**&#51060;&#46972;&#44256; &#54620;&#45796;. &#50696;&#52769;&#44592;&#47560;&#45796; &#45224;&#44200;&#51652; 37%&#45716; &#47784;&#46160; &#45796;&#47476;&#45796;.\n",
    "\n",
    "&#46384;&#46972;&#49436; <U>&#50696;&#52769;&#44592;&#44032; &#54984;&#47144;&#46104;&#45716; &#46041;&#50504;&#50640;&#45716; oob &#49368;&#54540;&#51012; &#49324;&#50857;&#54616;&#51648; &#50506;&#51004;&#48064;&#47196; &#44160;&#51613; &#49464;&#53944;&#45208; &#44368;&#52264; &#44160;&#51613;&#51012; &#49324;&#50857;&#54616;&#51648; &#50506;&#44256; oob &#49368;&#54540;&#51012; &#49324;&#50857;&#54616;&#50668; &#54217;&#44032;&#54624; &#49688; &#51080;&#45796;</U>. &#50521;&#49345;&#48660;&#51032; &#54217;&#44032;&#45716; &#44033; &#50696;&#52769;&#44592;&#51032; oob &#54217;&#44032;&#47484; &#54217;&#44512;&#54616;&#50668; &#50619;&#45716;&#45796;.\n",
    "\n",
    "Sklearn&#50640;&#49436;&#45716; `oob_score_`&#47484; &#51060;&#50857;&#54620;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bag_clf = BaggingClassifier(\n",
    "    DecisionTreeClassifier(random_state=42), n_estimators=500,\n",
    "    bootstrap=True, n_jobs=-1, oob_score=True\n",
    ")\n",
    "\n",
    "bag_clf.fit(train_x, train_y)\n",
    "bag_clf.oob_score_  # oob&#47484; &#51060;&#50857;&#54620; &#44160;&#51613; &#51216;&#49688;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.912"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_oob = bag_clf.predict(test_x)\n",
    "accuracy_score(test_y, pred_oob)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.1746988 , 0.8253012 ],\n",
       "       [0.34554974, 0.65445026],\n",
       "       [1.        , 0.        ],\n",
       "       ...,\n",
       "       [0.98395722, 0.01604278],\n",
       "       [0.82828283, 0.17171717],\n",
       "       [0.        , 1.        ]])"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# oob sample&#50640; &#45824;&#54620; &#44208;&#44284;&#44050;&#51012; &#48380; &#49688; &#51080;&#45796;.\n",
    "bag_clf.oob_decision_function_ "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Random patch&#50752; random subspace\n",
    "- Random patch : &#54984;&#47144; &#53945;&#49457;&#44284; &#49368;&#54540;&#51012; &#47784;&#46160; &#49368;&#54540;&#47553;&#54616;&#45716; &#44163; \n",
    " - `bootstrap_features=False`, `max_samples=1.0`\n",
    "- Random subspace : &#54984;&#47144; &#53945;&#49457;&#51008; &#49368;&#54540;&#47553;&#54616;&#44256; &#54984;&#47144; &#49368;&#54540;&#51012; &#47784;&#46160; &#49324;&#50857;&#54616;&#45716; &#44163; \n",
    " - `bootstrap_features=True`, `max_samples < 1.0`"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3. Random forest\n",
    "- &#47004;&#45924; &#54252;&#47112;&#49828;&#53944;&#45716; &#51068;&#48152;&#51201;&#51004;&#47196; bagging &#48169;&#48277;(&#46608;&#45716; pasting)&#51012; &#51201;&#50857;&#54620; &#44208;&#51221; &#53944;&#47532;&#51032; &#50521;&#49345;&#48660;&#51060;&#45796;.\n",
    "- &#47004;&#45924; &#54252;&#47112;&#49828;&#53944; &#50508;&#44256;&#47532;&#51608;&#51008; **&#53944;&#47532;&#51032; &#45432;&#46300;&#47484; &#48516;&#54624;&#54624; &#46412; &#51204;&#52404; &#53945;&#49457; &#51473;&#50640;&#49436; &#52572;&#49440;&#51032; &#53945;&#49457;&#51012; &#52286;&#45716; &#45824;&#49888; <font color=red>&#47924;&#51089;&#50948;&#47196; &#49440;&#53469;&#54620; &#53945;&#49457; &#54980;&#48372; &#51473;&#50640;&#49436; &#52572;&#51201;&#51032; &#53945;&#49457;&#51012; &#52286;&#45716;</font> &#49885;&#51004;&#47196; &#47924;&#51089;&#50948;&#49457;&#51012; &#45908; &#51452;&#51077;**&#54620;&#45796;.\n",
    " - &#53944;&#47532;&#47484; &#45908;&#50865; &#45796;&#50577;&#54616;&#44172; &#49373;&#49457; (&#53944;&#47532;&#51032; &#51032;&#51316;&#49457;&#51012; &#45230;&#52628;&#44256;, &#45796;&#50577;&#49457;&#51012; &#51613;&#44032;)\n",
    " - &#54200;&#54693;&#51012; &#49552;&#54644; &#48372;&#45716; &#45824;&#49888; &#48516;&#49328;&#51012; &#45230;&#52628;&#50612; &#51204;&#52404;&#51201;&#51004;&#47196; &#45908; &#54988;&#47469;&#54620; &#47784;&#45944;&#51012; &#49373;&#49457;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Sklearn&#50640;&#49436; &#51228;&#44277;&#54616;&#45716; `RandomForestClassifier`&#51012; &#49324;&#50857;&#54616;&#47732; &#46108;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features='auto', max_leaf_nodes=16,\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, n_estimators=500, n_jobs=-1,\n",
       "            oob_score=False, random_state=42, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "# 16&#44060;&#51032; &#47532;&#54532; &#45432;&#46300;&#47484; &#44054;&#45716; 500&#44060;&#51032; &#53944;&#47532;&#47196; &#51060;&#47336;&#50612;&#51652; &#47004;&#45924; &#54252;&#47112;&#49828;&#53944;&#47484; &#50668;&#47084; CPU &#53076;&#50612;&#50640;&#49436; &#54984;&#47144;\n",
    "rnd_clf = RandomForestClassifier(n_estimators=500, max_leaf_nodes=16, n_jobs=-1, random_state=42)\n",
    "rnd_clf.fit(train_x, train_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.92"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_rf_y = rnd_clf.predict(test_x)\n",
    "accuracy_score(test_y, pred_rf_y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Sklearn&#50640;&#49436; &#51228;&#44277;&#54616;&#45716; `BaggingClassifier` &#51060;&#50857;&#54616;&#50668; Random forest&#47484; &#44396;&#54788;&#54616;&#45716; &#48169;&#48277;&#46020; &#51080;&#45796;. `splitter=\"random\"`&#51012; &#49444;&#51221;&#54616;&#47732; &#46121;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BaggingClassifier(base_estimator=DecisionTreeClassifier(class_weight=None, criterion='gini', max_depth=None,\n",
       "            max_features=None, max_leaf_nodes=16,\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=False, random_state=42,\n",
       "            splitter='random'),\n",
       "         bootstrap=True, bootstrap_features=False, max_features=1.0,\n",
       "         max_samples=1.0, n_estimators=500, n_jobs=-1, oob_score=False,\n",
       "         random_state=42, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bag_clf = BaggingClassifier(\n",
    "    DecisionTreeClassifier(splitter=\"random\", max_leaf_nodes=16, random_state=42),\n",
    "    n_estimators=500, \n",
    "    max_samples=1.0, \n",
    "    bootstrap=True, \n",
    "    n_jobs=-1, \n",
    "    random_state=42)\n",
    "\n",
    "bag_clf.fit(train_x, train_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.92"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred = bag_clf.predict(test_x)\n",
    "accuracy_score(test_y, pred_rf_y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Extra tree"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Extremely Randomized Trees (Extra Trees) : &#47004;&#45924; &#54252;&#47112;&#49828;&#53944;&#50640;&#49436; &#53944;&#47532;&#47484; &#47564;&#46308; &#46412; &#44033; &#45432;&#46300;&#45716; &#47924;&#51089;&#50948;&#47196; &#53945;&#49457;&#51032; &#49436;&#48652;&#49483;&#51012; &#47564;&#46308;&#50612; &#48516;&#54624;&#50640; &#49324;&#50857;&#54620;&#45796;. &#53945;&#49457;&#51032; &#49436;&#48652;&#49483;&#50640; &#45824;&#54616;&#50668;, &#44033; &#53945;&#49457; &#47784;&#46160;&#50640; &#45824;&#54644;&#49436; &#44228;&#49328;&#51012; &#54616;&#44256; &#48516;&#54624;&#50640; &#52572;&#51201;&#54868;&#46108; &#51076;&#44228;&#44050;&#51012; &#52286;&#45716;&#45796;(greedy search). **&#53944;&#47532;&#47484; &#45908;&#50865; &#47924;&#51089;&#50948;&#47196; &#47564;&#46308;&#44592; &#50948;&#54644;**, &#48372;&#53685;&#51032; &#44208;&#51221; &#53944;&#47532;&#52376;&#47100; &#54980;&#48372; &#53945;&#49457;&#51032; greedy search&#47484; &#54616;&#50668; &#52572;&#51201;&#51032; &#51076;&#44247;&#44050;&#51012; &#52286;&#45716; &#45824;&#49888; **&#54980;&#48372; &#53945;&#49457;&#51012; &#49324;&#50857;&#54644; &#47924;&#51089;&#50948;&#47196; &#48516;&#54624;**&#54620; &#45796;&#51020; &#44536;&#51473;&#50640;&#49436; &#52572;&#49345;&#51032; &#48516;&#54624;&#51012; &#49440;&#53469;&#54620;&#45796;. &#46384;&#46972;&#49436; &#51068;&#48152;&#51201;&#51064; &#47004;&#45924; &#54252;&#47112;&#49828;&#53944;&#48372;&#45796; &#50641;&#49828;&#53944;&#46972; &#53944;&#47532;&#44032; &#55016;&#50476; &#48736;&#47476;&#45796;.\n",
    "- `RandomForestClassifier` vs `ExtraTreesClassifier` : &#50612;&#46500; &#44163;&#51060; &#45908; &#51339;&#51012;&#51648;&#45716; &#49549;&#45800;&#54616;&#44592; &#50612;&#47157;&#45796;. &#51068;&#48152;&#51201;&#51004;&#47196; &#46168; &#45796; &#49884;&#46020;&#54644;&#48372;&#44256; &#44368;&#52264; &#44160;&#51613;&#51004;&#47196; &#48708;&#44368;&#54644;&#48372;&#45716; &#44163;&#51060; &#50976;&#51068;&#54620; &#48169;&#48277;&#51060;&#45796;. &#44536;&#47532;&#44256; &#44536;&#47532;&#46300; &#53456;&#49353;&#51004;&#47196; &#54616;&#51060;&#54140;&#54028;&#46972;&#48120;&#53552; &#53916;&#45789;&#51012; &#54620;&#45796;."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### &#53945;&#49457; &#51473;&#50836;&#46020;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- &#47004;&#45924; &#54252;&#47112;&#49828;&#53944;&#51032; &#46608; &#45796;&#47480; &#51109;&#51216;&#51008; &#53945;&#49457;&#51032; &#49345;&#45824;&#51201; &#51473;&#50836;&#46020;&#47484; &#52769;&#51221;&#54616;&#44592; &#49789;&#45796;&#45716; &#44163;&#51060;&#45796;.\n",
    "- Sklearn&#51008; &#50612;&#46500; &#53945;&#49457;&#51012; &#49324;&#50857;&#54620; &#45432;&#46300;&#44032; (&#47004;&#45924; &#54252;&#47112;&#49828;&#53944;&#50640; &#51080;&#45716; &#47784;&#46304; &#53944;&#47532;&#50640; &#44152;&#52432;&#49436;) &#54217;&#44512;&#51201;&#51004;&#47196; &#48520;&#49692;&#46020;&#47484; &#50620;&#47560;&#45208; &#44048;&#49548;&#49884;&#53412;&#45716;&#51648; &#54869;&#51064;&#54616;&#50668; &#53945;&#49457;&#51032; &#51473;&#50836;&#46020;&#47484; &#52769;&#51221;&#54620;&#45796;.\n",
    " - &#44032;&#51473;&#52824;&#51032; &#54217;&#44512;&#51060;&#47728; &#44033; &#45432;&#46300;&#51032; &#44032;&#51473;&#52824;&#45716; &#50672;&#44288;&#46108; &#54984;&#47144; &#49368;&#54540; &#49688;&#50752; &#44057;&#45796;.\n",
    " - &#44208;&#51221; &#53944;&#47532;&#51032; &#53945;&#49457; &#51473;&#50836;&#46020;&#45716; &#45432;&#46300;&#50640; &#49324;&#50857;&#46108; &#53945;&#49457;&#48324; information gain&#50640;&#49436; &#49368;&#54540; &#49688;&#47564; &#49368;&#54540; &#48708;&#50984;&#47196; &#48148;&#44984;&#50612;,\n",
    " $$(\\text{&#54788;&#51116; &#45432;&#46300;&#51032; &#49368;&#54540; &#48708;&#50984;} \\times \\text{&#48520;&#49692;&#46020;}) - (\\text{&#51901; &#51088;&#49885; &#45432;&#46300;&#51032; &#49368;&#54540; &#48708;&#50984;} \\times \\text{&#48520;&#49692;&#46020;}) - (\\text{&#50724;&#47480;&#51901; &#45432;&#46300;&#51032; &#49368;&#54540; &#48708;&#50984;} \\times \\text{&#48520;&#49692;&#46020;})$$\n",
    " &#51012; &#44228;&#49328;&#54616;&#50668; &#45908;&#54616;&#44256;, &#53945;&#49457; &#51473;&#50836;&#46020;&#51032; &#54633;&#51060; 1&#51060; &#46104;&#46020;&#47197; &#51204;&#52404; &#54633;&#51004;&#47196; &#45208;&#45572;&#50612; &#51221;&#44508;&#54868;&#54633;&#45768;&#45796;. &#50668;&#44592;&#49436; &#49368;&#54540; &#48708;&#50984;&#51008; &#53944;&#47532; &#51204;&#52404; &#49368;&#54540; &#49688;&#50640; &#45824;&#54620; &#48708;&#50984;&#51077;&#45768;&#45796;. &#47004;&#45924; &#54252;&#47112;&#49828;&#53944;&#51032; &#53945;&#49457; &#51473;&#50836;&#46020;&#45716; &#44033; &#44208;&#51221; &#53944;&#47532;&#51032; &#53945;&#51221; &#51473;&#50836;&#46020;&#47484; &#47784;&#46160; &#44228;&#49328;&#54616;&#50668; &#45908;&#54620; &#54980; &#53944;&#47532; &#49688;&#47196; &#45208;&#45576; &#44163;&#51077;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.43931025, 0.56068975])"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rnd_clf.feature_importances_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Iris&#47484; random forest&#47484; &#51060;&#50857;&#54616;&#50668; &#48516;&#47448;&#54616;&#44256; &#45212; &#54980;, feature &#51473;&#50836;&#46020;&#47484; &#44396;&#54644;&#48372;&#51088;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sepal length (cm) 0.11249225099876374\n",
      "sepal width (cm) 0.023119288282510326\n",
      "petal length (cm) 0.44103046436395765\n",
      "petal width (cm) 0.4233579963547681\n"
     ]
    }
   ],
   "source": [
    "# Iris data set\n",
    "\n",
    "from sklearn.datasets import load_iris\n",
    "iris = load_iris()\n",
    "rnd_clf = RandomForestClassifier(n_estimators=500, n_jobs=-1, random_state=42)\n",
    "rnd_clf.fit(iris[\"data\"], iris[\"target\"])\n",
    "for name, score in zip(iris[\"feature_names\"], rnd_clf.feature_importances_):\n",
    "    print(name, score)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "&#49884;&#44033;&#54868;&#47484; &#54644;&#48372;&#51088;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5,0,'Relative Importance')"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a1de6ceb8>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "rf_feat_importance = rnd_clf.feature_importances_\n",
    "indices = np.argsort(rf_feat_importance)\n",
    "\n",
    "plt.title('RF feature importnace')\n",
    "plt.barh(range(len(indices)), rf_feat_importance[indices], color='b', align='center')\n",
    "plt.yticks(range(len(indices)), np.array(iris[\"feature_names\"])[indices])\n",
    "plt.xlabel('Relative Importance')"
   ]
  }
 ],
 "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": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}