06_Объединение алгоритмов в цепочки и конвейеры.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true,
    "hide_input": false
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import mglearn\n",
    "%matplotlib inline\n",
    "from IPython.display import display\n",
    "plt.rc('font', family='Verdana')\n",
    "\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.datasets import load_breast_cancer\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "\n",
    "# загружаем и разбиваем данные\n",
    "cancer = load_breast_cancer()\n",
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    cancer.data, cancer.target, random_state=0)\n",
    "\n",
    "# вычисляем минимум и максимум по обучающим данным\n",
    "scaler = MinMaxScaler().fit(X_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Цепочки и конвейеры"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Правильность на тестовом наборе: 0.95\n"
     ]
    }
   ],
   "source": [
    "# масштабируем обучающие данные\n",
    "X_train_scaled = scaler.transform(X_train)\n",
    "\n",
    "svm = SVC()\n",
    "# строим SVM на масштабированных обучающих данных\n",
    "svm.fit(X_train_scaled, y_train)\n",
    "# масштабируем тестовые данные и оцениваем качество на масштабированных данных\n",
    "X_test_scaled = scaler.transform(X_test)\n",
    "print(\"Правильность на тестовом наборе: {:.2f}\".format(svm.score(X_test_scaled, y_test)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Отбор параметров с помощью предварительной обработки"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Наил знач правильности перекр проверки: 0.98\n",
      "Наил знач правильности на тесте: 0.97\n",
      "Наил параметры:  {'C': 1, 'gamma': 1}\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "# только в иллюстративных целях, не используйте этот код!\n",
    "param_grid = {'C': [0.001, 0.01, 0.1, 1, 10, 100],\n",
    "'gamma': [0.001, 0.01, 0.1, 1, 10, 100]}\n",
    "grid = GridSearchCV(SVC(), param_grid=param_grid, cv=5)\n",
    "grid.fit(X_train_scaled, y_train)\n",
    "print(\"Наил знач правильности перекр проверки: {:.2f}\".format(grid.best_score_))\n",
    "print(\"Наил знач правильности на тесте: {:.2f}\".format(grid.score(X_test_scaled, y_test)))\n",
    "print(\"Наил параметры: \", grid.best_params_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "hide_input": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA1cAAAJICAYAAACNNR4KAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nPKC4AgAAAAAPKK4AAAAAwIP/Dy5BmDkwRdbTAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1eaff5eeb70>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "mglearn.plots.plot_improper_processing()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Построение конвейеров"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.pipeline import Pipeline\n",
    "pipe = Pipeline([(\"scaler\", MinMaxScaler()), (\"svm\", SVC())])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Pipeline(memory=None,\n",
       "     steps=[('scaler', MinMaxScaler(copy=True, feature_range=(0, 1))), ('svm', SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0,\n",
       "  decision_function_shape='ovr', degree=3, gamma='auto', kernel='rbf',\n",
       "  max_iter=-1, probability=False, random_state=None, shrinking=True,\n",
       "  tol=0.001, verbose=False))])"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pipe.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Правильность на тестовом наборе: 0.95\n"
     ]
    }
   ],
   "source": [
    "print(\"Правильность на тестовом наборе: {:.2f}\".format(pipe.score(X_test, y_test)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Использование конвейера, помещенного в объект GridSearchCV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "param_grid = {'svm__C': [0.001, 0.01, 0.1, 1, 10, 100],\n",
    "              'svm__gamma': [0.001, 0.01, 0.1, 1, 10, 100]}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Наил значение правильности перекр проверки: 0.98\n",
      "Правильность на тестовом наборе: 0.97\n",
      "Наилучшие параметры: {'svm__C': 1, 'svm__gamma': 1}\n"
     ]
    }
   ],
   "source": [
    "grid = GridSearchCV(pipe, param_grid=param_grid, cv=5)\n",
    "grid.fit(X_train, y_train)\n",
    "print(\"Наил значение правильности перекр проверки: {:.2f}\".format(grid.best_score_))\n",
    "print(\"Правильность на тестовом наборе: {:.2f}\".format(grid.score(X_test, y_test)))\n",
    "print(\"Наилучшие параметры: {}\".format(grid.best_params_))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "hide_input": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA1cAAAHbCAYAAAAnL2B6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1eaff5ee9b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "mglearn.plots.plot_proper_processing()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "rnd = np.random.RandomState(seed=0)\n",
    "X = rnd.normal(size=(100, 10000))\n",
    "y = rnd.normal(size=(100,))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "форма массива X_selected: (100, 500)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.feature_selection import SelectPercentile, f_regression\n",
    "\n",
    "select = SelectPercentile(score_func=f_regression, percentile=5).fit(X, y)\n",
    "X_selected = select.transform(X)\n",
    "print(\"форма массива X_selected: {}\".format(X_selected.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "R-квадрат перекр проверки (cv только для ridge): 0.91\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.linear_model import Ridge\n",
    "print(\"R-квадрат перекр проверки (cv только для ridge): {:.2f}\".format(\n",
    "      np.mean(cross_val_score(Ridge(), X_selected, y, cv=5))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "R-квадрат перекр проверки (конвейер): -0.25\n"
     ]
    }
   ],
   "source": [
    "pipe = Pipeline([(\"select\", SelectPercentile(score_func=f_regression,\n",
    "                                             percentile=5)),\n",
    "                 (\"ridge\", Ridge())])\n",
    "print(\"R-квадрат перекр проверки (конвейер): {:.2f}\".format(\n",
    "np.mean(cross_val_score(pipe, X, y, cv=5))))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Общий интерфейс конвейера"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def fit(self, X, y):\n",
    "    X_transformed = X\n",
    "    for name, estimator in self.steps[:-1]:\n",
    "        # перебираем все этапы, кроме последнего\n",
    "        # подгоняем и преобразуем данные\n",
    "        X_transformed = estimator.fit_transform(X_transformed, y)\n",
    "    # осуществляем подгонку на последнем этапе\n",
    "    self.steps[-1][1].fit(X_transformed, y)\n",
    "    return self"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def predict(self, X):\n",
    "    X_transformed = X\n",
    "    for step in self.steps[:-1]:\n",
    "        # перебираем все этапы, кроме последнего\n",
    "        # преобразуем данные\n",
    "        X_transformed = step[1].transform(X_transformed)\n",
    "    # получаем прогнозы на последнем этапе\n",
    "    return self.steps[-1][1].predict(X_transformed)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![pipeline_illustration](images/pipeline.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Удобный способ построения конвейеров с помощью функции ``make_pipeline``"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.pipeline import make_pipeline\n",
    "# стандартный синтаксис\n",
    "pipe_long = Pipeline([(\"scaler\", MinMaxScaler()), (\"svm\", SVC(C=100))])\n",
    "# сокращенный синтаксис\n",
    "pipe_short = make_pipeline(MinMaxScaler(), SVC(C=100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Этапы конвейера:\n",
      "[('minmaxscaler', MinMaxScaler(copy=True, feature_range=(0, 1))), ('svc', SVC(C=100, cache_size=200, class_weight=None, coef0=0.0,\n",
      "  decision_function_shape='ovr', degree=3, gamma='auto', kernel='rbf',\n",
      "  max_iter=-1, probability=False, random_state=None, shrinking=True,\n",
      "  tol=0.001, verbose=False))]\n"
     ]
    }
   ],
   "source": [
    "print(\"Этапы конвейера:\\n{}\".format(pipe_short.steps))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Этапы конвейера:\n",
      "[('standardscaler-1', StandardScaler(copy=True, with_mean=True, with_std=True)), ('pca', PCA(copy=True, iterated_power='auto', n_components=2, random_state=None,\n",
      "  svd_solver='auto', tol=0.0, whiten=False)), ('standardscaler-2', StandardScaler(copy=True, with_mean=True, with_std=True))]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.decomposition import PCA\n",
    "\n",
    "pipe = make_pipeline(StandardScaler(), PCA(n_components=2), StandardScaler())\n",
    "print(\"Этапы конвейера:\\n{}\".format(pipe.steps))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Работа с атрибутами этапов"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "форма components: (2, 30)\n"
     ]
    }
   ],
   "source": [
    "# подгоняем заранее заданный конвейер к набору данных cancer\n",
    "pipe.fit(cancer.data)\n",
    "# извлекаем первые две главные компоненты на этапе \"pca\"\n",
    "components = pipe.named_steps[\"pca\"].components_\n",
    "print(\"форма components: {}\".format(components.shape))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Работа с атрибутами конвейера, помещенного в объект GridSearchCV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "pipe = make_pipeline(StandardScaler(), LogisticRegression())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "param_grid = {'logisticregression__C': [0.01, 0.1, 1, 10, 100]}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=5, error_score='raise',\n",
       "       estimator=Pipeline(memory=None,\n",
       "     steps=[('standardscaler', StandardScaler(copy=True, with_mean=True, with_std=True)), ('logisticregression', LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "          intercept_scaling=1, max_iter=100, multi_class='ovr', n_jobs=1,\n",
       "          penalty='l2', random_state=None, solver='liblinear', tol=0.0001,\n",
       "          verbose=0, warm_start=False))]),\n",
       "       fit_params=None, iid=True, n_jobs=1,\n",
       "       param_grid={'logisticregression__C': [0.01, 0.1, 1, 10, 100]},\n",
       "       pre_dispatch='2*n_jobs', refit=True, return_train_score='warn',\n",
       "       scoring=None, verbose=0)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    cancer.data, cancer.target, random_state=4)\n",
    "grid = GridSearchCV(pipe, param_grid, cv=5)\n",
    "grid.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Лучшая модель:\n",
      "Pipeline(memory=None,\n",
      "     steps=[('standardscaler', StandardScaler(copy=True, with_mean=True, with_std=True)), ('logisticregression', LogisticRegression(C=0.1, class_weight=None, dual=False, fit_intercept=True,\n",
      "          intercept_scaling=1, max_iter=100, multi_class='ovr', n_jobs=1,\n",
      "          penalty='l2', random_state=None, solver='liblinear', tol=0.0001,\n",
      "          verbose=0, warm_start=False))])\n"
     ]
    }
   ],
   "source": [
    "print(\"Лучшая модель:\\n{}\".format(grid.best_estimator_))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Этап логистической регрессии:\n",
      "LogisticRegression(C=0.1, class_weight=None, dual=False, fit_intercept=True,\n",
      "          intercept_scaling=1, max_iter=100, multi_class='ovr', n_jobs=1,\n",
      "          penalty='l2', random_state=None, solver='liblinear', tol=0.0001,\n",
      "          verbose=0, warm_start=False)\n"
     ]
    }
   ],
   "source": [
    "print(\"Этап логистической регрессии:\\n{}\".format(\n",
    "      grid.best_estimator_.named_steps[\"logisticregression\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Коэффициенты логистической регрессии:\n",
      "[[-0.38856355 -0.37529972 -0.37624793 -0.39649439 -0.11519359  0.01709608\n",
      "  -0.3550729  -0.38995414 -0.05780518  0.20879795 -0.49487753 -0.0036321\n",
      "  -0.37122718 -0.38337777 -0.04488715  0.19752816  0.00424822 -0.04857196\n",
      "   0.21023226  0.22444999 -0.54669761 -0.52542026 -0.49881157 -0.51451071\n",
      "  -0.39256847 -0.12293451 -0.38827425 -0.4169485  -0.32533663 -0.13926972]]\n"
     ]
    }
   ],
   "source": [
    "print(\"Коэффициенты логистической регрессии:\\n{}\".format(\n",
    "      grid.best_estimator_.named_steps[\"logisticregression\"].coef_))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Находим оптимальные параметры этапов конвейера с помощью решетчатого поиска"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.datasets import load_boston\n",
    "boston = load_boston()\n",
    "X_train, X_test, y_train, y_test = train_test_split(boston.data, boston.target,\n",
    "                                                    random_state=0)\n",
    "\n",
    "from sklearn.preprocessing import PolynomialFeatures\n",
    "pipe = make_pipeline(\n",
    "    StandardScaler(),\n",
    "    PolynomialFeatures(),\n",
    "    Ridge())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "param_grid = {'polynomialfeatures__degree': [1, 2, 3],\n",
    "              'ridge__alpha': [0.001, 0.01, 0.1, 1, 10, 100]}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=5, error_score='raise',\n",
       "       estimator=Pipeline(memory=None,\n",
       "     steps=[('standardscaler', StandardScaler(copy=True, with_mean=True, with_std=True)), ('polynomialfeatures', PolynomialFeatures(degree=2, include_bias=True, interaction_only=False)), ('ridge', Ridge(alpha=1.0, copy_X=True, fit_intercept=True, max_iter=None,\n",
       "   normalize=False, random_state=None, solver='auto', tol=0.001))]),\n",
       "       fit_params=None, iid=True, n_jobs=-1,\n",
       "       param_grid={'polynomialfeatures__degree': [1, 2, 3], 'ridge__alpha': [0.001, 0.01, 0.1, 1, 10, 100]},\n",
       "       pre_dispatch='2*n_jobs', refit=True, return_train_score='warn',\n",
       "       scoring=None, verbose=0)"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grid = GridSearchCV(pipe, param_grid=param_grid, cv=5, n_jobs=-1)\n",
    "grid.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.colorbar.Colorbar at 0x1eaff2d3940>"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAccAAAD3CAYAAABo8Z+rAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1ea846e24a8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.matshow(grid.cv_results_['mean_test_score'].reshape(3, -1),\n",
    "            vmin=0, cmap=\"viridis\")\n",
    "plt.xlabel(\"ridge__alpha\")\n",
    "plt.ylabel(\"polynomialfeatures__degree\")\n",
    "plt.xticks(range(len(param_grid['ridge__alpha'])), param_grid['ridge__alpha'])\n",
    "plt.yticks(range(len(param_grid['polynomialfeatures__degree'])),\n",
    "           param_grid['polynomialfeatures__degree'])\n",
    "\n",
    "plt.colorbar()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Наилучшие параметры: {'polynomialfeatures__degree': 2, 'ridge__alpha': 10}\n"
     ]
    }
   ],
   "source": [
    "print(\"Наилучшие параметры: {}\".format(grid.best_params_))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "R-квадрат на тестовом наборе: 0.77\n"
     ]
    }
   ],
   "source": [
    "print(\"R-квадрат на тестовом наборе: {:.2f}\".format(grid.score(X_test, y_test)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "R-квадрат без полином. преобразования: 0.63\n"
     ]
    }
   ],
   "source": [
    "param_grid = {'ridge__alpha': [0.001, 0.01, 0.1, 1, 10, 100]}\n",
    "pipe = make_pipeline(StandardScaler(), Ridge())\n",
    "grid = GridSearchCV(pipe, param_grid, cv=5)\n",
    "grid.fit(X_train, y_train)\n",
    "print(\"R-квадрат без полином. преобразования: {:.2f}\".format(grid.score(X_test, y_test)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "pipe = Pipeline([('preprocessing', StandardScaler()), ('classifier', SVC())])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "param_grid = [\n",
    "    {'classifier': [SVC()], 'preprocessing': [StandardScaler(), None],\n",
    "     'classifier__gamma': [0.001, 0.01, 0.1, 1, 10, 100],\n",
    "     'classifier__C': [0.001, 0.01, 0.1, 1, 10, 100]},\n",
    "    {'classifier': [RandomForestClassifier(n_estimators=100)],\n",
    "     'preprocessing': [None], 'classifier__max_features': [1, 2, 3]}]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Наилучшие параметры:\n",
      "{'classifier': SVC(C=10, cache_size=200, class_weight=None, coef0=0.0,\n",
      "  decision_function_shape='ovr', degree=3, gamma=0.01, kernel='rbf',\n",
      "  max_iter=-1, probability=False, random_state=None, shrinking=True,\n",
      "  tol=0.001, verbose=False), 'classifier__C': 10, 'classifier__gamma': 0.01, 'preprocessing': StandardScaler(copy=True, with_mean=True, with_std=True)}\n",
      "\n",
      "Наил значение правильности перекр проверки: 0.99\n",
      "Правильность на тестовом наборе: 0.98\n"
     ]
    }
   ],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    cancer.data, cancer.target, random_state=0)\n",
    "\n",
    "grid = GridSearchCV(pipe, param_grid, cv=5)\n",
    "grid.fit(X_train, y_train)\n",
    "\n",
    "print(\"Наилучшие параметры:\\n{}\\n\".format(grid.best_params_))\n",
    "print(\"Наил значение правильности перекр проверки: {:.2f}\".format(grid.best_score_))\n",
    "print(\"Правильность на тестовом наборе: {:.2f}\".format(grid.score(X_test, y_test)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Выводы и перспективы"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}