HomeWork5.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Домашнее задание №5\n",
    "\n",
    "Data Set Information:\n",
    "\n",
    "The dataset was acquired and annotated by professional physicians at 'Hospital Universitario de Caracas'.\n",
    "The subjective judgments (target variables) were originally done in an ordinal manner (poor, fair, good, excellent) and was discretized in two classes (bad, good).\n",
    "Images were randomly sampled from the original colposcopic sequences (videos).\n",
    "The original images and the manual segmentations are included in the 'images' directory.\n",
    "The dataset has three modalities (i.e. Hinselmann, Green, Schiller).\n",
    "The target variables are expert::X (X in 0,...,5) and consensus.\n",
    "cervix_area: image area with cervix.\n",
    "\n",
    "os_area: image area with external os.\n",
    "\n",
    "walls_area: image area with vaginal walls.\n",
    "\n",
    "speculum_area: image area with the speculum.\n",
    "\n",
    "artifacts_area: image area with artifacts.\n",
    "\n",
    "cervix_artifacts_area: cervix area with the artifacts.\n",
    "\n",
    "os_artifacts_area: external os area with the artifacts.\n",
    "\n",
    "walls_artifacts_area: vaginal walls with the artifacts.\n",
    "\n",
    "speculum_artifacts_area: speculum area with the artifacts.\n",
    "\n",
    "cervix_specularities_area: cervix area with the specular reflections.\n",
    "\n",
    "os_specularities_area: external os area with the specular reflections.\n",
    "\n",
    "walls_specularities_area: vaginal walls area with the specular reflections.\n",
    "\n",
    "speculum_specularities_area: speculum area with the specular reflections.\n",
    "\n",
    "specularities_area: total area with specular reflections.\n",
    "\n",
    "area_h_max_diff: maximum area differences between the four cervix quadrants.\n",
    "\n",
    "rgb_cervix_r_mean: average color information in the cervix (R channel).\n",
    "\n",
    "rgb_cervix_r_std: stddev color information in the cervix (R channel).\n",
    "\n",
    "rgb_cervix_r_mean_minus_std: (avg - stddev) color information in the cervix (R channel).\n",
    "\n",
    "rgb_cervix_r_mean_plus_std: (avg + stddev) information in the cervix (R channel).\n",
    "\n",
    "rgb_cervix_g_mean: average color information in the cervix (G channel).\n",
    "\n",
    "rgb_cervix_g_std: stddev color information in the cervix (G channel).\n",
    "\n",
    "rgb_cervix_g_mean_minus_std: (avg - stddev) color information in the cervix (G channel).\n",
    "\n",
    "rgb_cervix_g_mean_plus_std: (avg + stddev) color information in the cervix (G channel).\n",
    "\n",
    "rgb_cervix_b_mean: average color information in the cervix (B channel).\n",
    "\n",
    "rgb_cervix_b_std: stddev color information in the cervix (B channel).\n",
    "\n",
    "rgb_cervix_b_mean_minus_std: (avg - stddev) color information in the cervix (B channel).\n",
    "\n",
    "rgb_cervix_b_mean_plus_std: (avg + stddev) color information in the cervix (B channel).\n",
    "\n",
    "rgb_total_r_mean: average color information in the image (B channel).\n",
    "\n",
    "rgb_total_r_std: stddev color information in the image (R channel).\n",
    "\n",
    "rgb_total_r_mean_minus_std: (avg - stddev) color information in the image (R channel).\n",
    "\n",
    "rgb_total_r_mean_plus_std: (avg + stddev) color information in the image (R channel).\n",
    "\n",
    "rgb_total_g_mean: average color information in the image (G channel).\n",
    "\n",
    "rgb_total_g_std: stddev color information in the image (G channel).\n",
    "\n",
    "rgb_total_g_mean_minus_std: (avg - stddev) color information in the image (G channel).\n",
    "\n",
    "rgb_total_g_mean_plus_std: (avg + stddev) color information in the image (G channel).\n",
    "\n",
    "rgb_total_b_mean: average color information in the image (B channel).\n",
    "\n",
    "rgb_total_b_std: stddev color information in the image (B channel).\n",
    "\n",
    "rgb_total_b_mean_minus_std: (avg - stddev) color information in the image (B channel).\n",
    "\n",
    "rgb_total_b_mean_plus_std: (avg + stddev) color information in the image (B channel).\n",
    "\n",
    "hsv_cervix_h_mean: average color information in the cervix (H channel).\n",
    "\n",
    "hsv_cervix_h_std: stddev color information in the cervix (H channel).\n",
    "\n",
    "hsv_cervix_s_mean: average color information in the cervix (S channel).\n",
    "\n",
    "hsv_cervix_s_std: stddev color information in the cervix (S channel).\n",
    "\n",
    "hsv_cervix_v_mean: average color information in the cervix (V channel).\n",
    "\n",
    "hsv_cervix_v_std: stddev color information in the cervix (V channel).\n",
    "\n",
    "hsv_total_h_mean: average color information in the image (H channel).\n",
    "\n",
    "hsv_total_h_std: stddev color information in the image (H channel).\n",
    "\n",
    "hsv_total_s_mean: average color information in the image (S channel).\n",
    "\n",
    "hsv_total_s_std: stddev color information in the image (S channel).\n",
    "\n",
    "hsv_total_v_mean: average color information in the image (V channel).\n",
    "\n",
    "hsv_total_v_std: stddev color information in the image (V channel).\n",
    "\n",
    "fit_cervix_hull_rate: Coverage of the cervix convex hull by the cervix.\n",
    "\n",
    "fit_cervix_hull_total: Image coverage of the cervix convex hull.\n",
    "\n",
    "fit_cervix_bbox_rate: Coverage of the cervix bounding box by the cervix.\n",
    "\n",
    "fit_cervix_bbox_total: Image coverage of the cervix bounding box.\n",
    "\n",
    "fit_circle_rate: Coverage of the cervix circle by the cervix.\n",
    "\n",
    "fit_circle_total: Image coverage of the cervix circle.\n",
    "\n",
    "fit_ellipse_rate: Coverage of the cervix ellipse by the cervix.\n",
    "\n",
    "fit_ellipse_total: Image coverage of the cervix ellipse.\n",
    "\n",
    "fit_ellipse_goodness: Goodness of the ellipse fitting.\n",
    "\n",
    "dist_to_center_cervix: Distance between the cervix center and the image center.\n",
    "\n",
    "dist_to_center_os: Distance between the cervical os center and the image center.\n",
    "\n",
    "experts::0: subjective assessment of the Expert 0 (target variable). experts::1: subjective assessment of the Expert 1 (target variable). experts::2: subjective assessment of the Expert 2 (target variable). experts::3: subjective assessment of the Expert 3 (target variable). experts::4: subjective assessment of the Expert 4 (target variable). experts::5: subjective assessment of the Expert 5 (target variable). consensus: subjective assessment of the consensus (target variable)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Считывание данных"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.read_csv('green.csv', sep = ',', engine = 'python')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "Задаем x и y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X = data.drop(('consensus'), axis = 1)\n",
    "y = data['consensus']\n",
    "\n",
    "from sklearn import preprocessing\n",
    "le = preprocessing.LabelEncoder()\n",
    "le.fit(y.unique())\n",
    "y_tr = le.transform(y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Разобьем выборку на тестовую и обучаемую"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.33, random_state = 7)\n",
    "N_test, _ = X_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Объявим массивы ошибок"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test_err_RF = []\n",
    "train_err_RF = []\n",
    "oob_err_RF = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "n_estimators = np.arange(1, 50, 1)\n",
    "\n",
    "for count in n_estimators:\n",
    "    rf_model = RandomForestClassifier(n_estimators = count, warm_start = True, oob_score = True)\n",
    "    rf_model.fit(X_train, y_train)\n",
    "    \n",
    "    y_train_pred = rf_model.predict(X_train)\n",
    "    y_test_pred = rf_model.predict(X_test)\n",
    "    \n",
    "    train_err_RF.append(np.mean(y_train != y_train_pred))\n",
    "    test_err_RF.append(np.mean(y_test != y_test_pred))\n",
    "    \n",
    "    oob_err_RF.append(1 - rf_model.oob_score_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0 0.151515151515\n"
     ]
    }
   ],
   "source": [
    "print(np.mean(y_train != y_train_pred), np.mean(y_test != y_test_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEXCAYAAABCjVgAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x26b5f3132e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(n_estimators, train_err_RF, 'b-o', label = 'train')\n",
    "plt.plot(n_estimators, test_err_RF, 'r-o', label = 'test')\n",
    "plt.plot(n_estimators, oob_err_RF, 'g-o', label = 'oob')\n",
    "plt.title('Error vs. n_estimators')\n",
    "plt.xlabel('n_estimators')\n",
    "plt.ylabel('error')\n",
    "plt.legend()\n",
    "pass"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Минимальные значения ошибок на обучающей и тестовой выборке"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0 0.0909090909091\n"
     ]
    }
   ],
   "source": [
    "min_train_err = np.min(train_err_RF)\n",
    "min_test_err_RF = np.min(test_err_RF)\n",
    "print(min_train_err, min_test_err_RF)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Недообучения не наблюдается, т. к. на обучающей выборке ошибка примерно равна нулю. Переобучение так же не наблюдается из- за отсутвия значительного роста ошибки, да и среднее ее значение не так велико"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "8\n"
     ]
    }
   ],
   "source": [
    "for i in range(len(test_err_RF)):\n",
    "    if((test_err_RF[i] == min_test_err_RF) and (train_err_RF[i] == min_train_err)):\n",
    "        n_estimator_opt = n_estimators[i]\n",
    "        break\n",
    "print(n_estimator_opt)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Для достижения подобного результата достаточно 8 деревьев "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Extremely Random Trees"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import ExtraTreesClassifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Объявляем массивы ошибок"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test_err_ERT = []\n",
    "train_err_ERT = []\n",
    "oob_err_ERT = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "for count in n_estimators:\n",
    "    ERT_model = ExtraTreesClassifier(n_estimators=count, bootstrap=True, oob_score = True)\n",
    "    ERT_model.fit(X_train, y_train)\n",
    "    \n",
    "    y_train_pred = ERT_model.predict(X_train)\n",
    "    y_test_pred = ERT_model.predict(X_test)\n",
    "    \n",
    "    train_err_ERT.append(np.mean(y_train != y_train_pred))\n",
    "    test_err_ERT.append(np.mean(y_test != y_test_pred))\n",
    "    \n",
    "    oob_err_ERT.append(1 - ERT_model.oob_score_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEXCAYAAABCjVgAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nVXQKIYQQqv4EVzn6ExATxoAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x26b5f5357b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(n_estimators, train_err_ERT, 'b-o', label = 'train')\n",
    "plt.plot(n_estimators, test_err_ERT, 'r-o', label = 'test')\n",
    "plt.plot(n_estimators, oob_err_ERT, 'g-o', label = 'oob')\n",
    "plt.title('Error vs. n_estimators')\n",
    "plt.xlabel('n_estimators')\n",
    "plt.ylabel('error')\n",
    "plt.legend()\n",
    "pass"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Минимальные значения ошибок на обучающей и тестовой выборке"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0 0.030303030303\n"
     ]
    }
   ],
   "source": [
    "min_train_err = np.min(train_err_ERT)\n",
    "min_test_err_ERT = np.min(test_err_ERT)\n",
    "print(min_train_err, min_test_err_ERT)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Недообучение не наблюдается, так же как и переобучение из- за малого значения ошибки на выборках и отсутствия значительного роста ошибки"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "31\n"
     ]
    }
   ],
   "source": [
    "for i in range(len(test_err_ERT)):\n",
    "    if((test_err_ERT[i] == min_test_err_ERT) and (train_err_ERT[i] == min_train_err)):\n",
    "        n_estimator_opt = n_estimators[i]\n",
    "        break\n",
    "print(n_estimator_opt)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Для достижения оптимального значения достаточно 31 дерева"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# AdaBoost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import AdaBoostClassifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Объявляем массивы ошибок "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test_err_AB = []\n",
    "train_err_AB = []\n",
    "oob_err_AB = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "for count in n_estimators:\n",
    "    AB_model = AdaBoostClassifier(n_estimators = count)\n",
    "    AB_model.fit(X_train, y_train)\n",
    "    \n",
    "    y_train_pred = AB_model.predict(X_train)\n",
    "    y_test_pred = AB_model.predict(X_test)\n",
    "    \n",
    "    train_err_AB.append(np.mean(y_train != y_train_pred))\n",
    "    test_err_AB.append(np.mean(y_test != y_test_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZIAAAEXCAYAAACH/8KRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x26b5f4ca5f8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(n_estimators, train_err_AB, 'b-o', label = 'train')\n",
    "plt.plot(n_estimators, test_err_AB, 'r-o', label = 'test')\n",
    "plt.title('Error vs. n_estimators')\n",
    "plt.xlabel('n_estimators')\n",
    "plt.ylabel('error')\n",
    "plt.legend()\n",
    "pass"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Минимальные значения ошибок на обучающей и тестовой выборке"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0 0.0606060606061\n"
     ]
    }
   ],
   "source": [
    "min_train_err = np.min(train_err_AB)\n",
    "min_test_err_AB = np.min(test_err_AB)\n",
    "print(min_train_err, min_test_err_AB)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Недообучение не наблюдается, поскольку величина ошибки на обучаемой выборке примерно равна нулю. Ошибка на тестовой выборке ведет себя скачкообразно, но переобучение не наблюдается"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4\n"
     ]
    }
   ],
   "source": [
    "for i in range(len(test_err_AB)):\n",
    "    if((test_err_AB[i] == min_test_err_AB) and (train_err_AB[i] == min_train_err)):\n",
    "        n_estimator_opt = n_estimators[i]\n",
    "        break\n",
    "print(n_estimator_opt)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Подобного результата можно достичь при количестве деревьев равном 4"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Gradient Boosted Trees"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import GradientBoostingClassifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " Объявляем массивы ошибок"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test_err_GBT = []\n",
    "train_err_GBT = []\n",
    "oob_err_GBT = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "for count in n_estimators:\n",
    "    GBT_model = GradientBoostingClassifier(n_estimators=count)\n",
    "    GBT_model.fit(X_train, y_train)\n",
    "    \n",
    "    y_train_pred = GBT_model.predict(X_train)\n",
    "    y_test_pred = GBT_model.predict(X_test)\n",
    "    \n",
    "    train_err_GBT.append(np.mean(y_train != y_train_pred))\n",
    "    test_err_GBT.append(np.mean(y_test != y_test_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEXCAYAAABCjVgAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x26b5f40a400>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(n_estimators, train_err_GBT, 'b-o', label = 'train')\n",
    "plt.plot(n_estimators, test_err_GBT, 'r-o', label = 'test')\n",
    "plt.title('Error vs. n_estimators')\n",
    "plt.xlabel('n_estimators')\n",
    "plt.ylabel('error')\n",
    "plt.legend()\n",
    "pass"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Минимальные значения ошибок на обучающей и тестовой выборке"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0 0.121212121212\n"
     ]
    }
   ],
   "source": [
    "min_train_err = np.min(train_err_GBT)\n",
    "min_test_err_GBT = np.min(test_err_GBT)\n",
    "print(min_train_err, min_test_err_GBT)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Недообучения как и в предыдущих случаях не наблюдается из- за нулевой ошибки на обучающей выборке. Ошибка на тестовой выборке снижается только после 20 деревьев и становится стабильной, не достигая нулевого значения. С такими показателями врят ли можно говорить о переобучении."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "20\n"
     ]
    }
   ],
   "source": [
    "for i in range(len(test_err_GBT)):\n",
    "    if((test_err_GBT[i] == min_test_err_GBT) and (train_err_GBT[i] == min_train_err)):\n",
    "        n_estimator_opt = n_estimators[i]\n",
    "        break\n",
    "print(n_estimator_opt)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Сравним графики ошибки на тестовой выборке"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEWCAYAAACXGLsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrLGwWCwWS7NYY2GxWCyWZvk/PAqUB2HRmQoAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x26b5f631e10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(n_estimators, test_err_RF, 'r-o', label = 'RF')\n",
    "plt.plot(n_estimators, test_err_ERT, 'b-o', label = 'ERT')\n",
    "plt.plot(n_estimators, test_err_AB, 'g-o', label = 'AB')\n",
    "plt.plot(n_estimators, test_err_GBT, 'y-o', label = 'GBT')\n",
    "plt.title('Error vs. Count of trees for test')\n",
    "plt.xlabel('Count of trees')\n",
    "plt.ylabel('Error')\n",
    "plt.legend()\n",
    "pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0909090909091 0.030303030303 0.0606060606061 0.121212121212\n"
     ]
    }
   ],
   "source": [
    "print(min_test_err_RF, min_test_err_ERT, min_test_err_AB, min_test_err_GBT)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "На основе полученных результатов нельзя выделить какой- то самый хороший алгоритм. Extremely Random Trees имеет самую малую ошибку, но довольно не стабилен исходя из графика. Gradient Boosted Trees напротив обретает стабильную ошибку после 20 деревьев, но эта ошибка довольно таки велика."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "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": "3.6.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}