CH2/templates/03_sklearn.linear_model_part1.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sklearn"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## sklearn.liner_model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**linear_model:**\n",
    "* RidgeClassifier\n",
    "* SGDClassifier\n",
    "* SGDRegressor\n",
    "* LinearRegression\n",
    "* LogisticRegression\n",
    "* Lasso\n",
    "* etc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "документация: http://scikit-learn.org/stable/modules/classes.html#module-sklearn.linear_model\n",
    "\n",
    "примеры: http://scikit-learn.org/stable/modules/linear_model.html#linear-model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from matplotlib.colors import ListedColormap\n",
    "from sklearn import cross_validation, datasets, linear_model, metrics\n",
    "\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Populating the interactive namespace from numpy and matplotlib\n"
     ]
    }
   ],
   "source": [
    "%pylab inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Генерация данных"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# создаем датасет из готового генератора\n",
    "blobs = datasets.make_blobs(centers = 2, cluster_std = 5.5, random_state=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7f04d9e4d610>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAewAAAHfCAYAAACSxOtjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAwa2JEk5YGBLkpQD/w+dR83o/VnFbAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f04dc2e0290>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "colors = ListedColormap(['red', 'blue'])\n",
    "\n",
    "pylab.figure(figsize(8, 8))\n",
    "pylab.scatter(map(lambda x: x[0], blobs[0]), map(lambda x: x[1], blobs[0]), c = blobs[1], cmap = colors)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Производим разовое разбиение данных на тренировочные данные и тестовые данные. cross_validation.train_test_split\n",
    "random_state = 1 - фиксация генератора для воспроизводимости "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "train_data, test_data, train_labels, test_labels = cross_validation.train_test_split(blobs[0], blobs[1], \n",
    "                                                                                    test_size = 0.3,\n",
    "                                                                                    random_state = 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Линейная классификация\n",
    " 1. При наличии 4х наборов данных <b>train_data, test_data, train_labels, test_labels, </b> проводим обучение классификатора на наборе <b>train_data</b>\n",
    " 2. Производим сравнение получившихся меток с метками <b>train_labels</b>\n",
    " 3. Оцениваем качество, и получаем коэффициенты свободного члена и признаков"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### RidgeClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#создание объекта - классификатора\n",
    "ridge_classifier = linear_model.RidgeClassifier(random_state = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RidgeClassifier(alpha=1.0, class_weight=None, copy_X=True, fit_intercept=True,\n",
       "        max_iter=None, normalize=False, random_state=1, solver='auto',\n",
       "        tol=0.001)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#обучение классификатора\n",
    "ridge_classifier.fit(train_data, train_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "#применение обученного классификатора\n",
    "ridge_predictions = ridge_classifier.predict(test_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0 0 0 1 0 0 0 0 0 1 0 1 0 1 1 1 0 0 0 1 0 0 0 1 1 1 1 0 0 1]\n"
     ]
    }
   ],
   "source": [
    "# получение меток тестовой модели\n",
    "print test_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0 0 0 1 0 1 0 0 0 1 0 1 0 1 1 1 1 0 1 1 1 0 0 1 1 1 1 0 0 1]\n"
     ]
    }
   ],
   "source": [
    "# получение меток обученной модели\n",
    "print ridge_predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8666666666666667"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#оценка качества классификации\n",
    "metrics.accuracy_score(test_labels, ridge_predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.0854443 , -0.07273219]])"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# получение весов признаков\n",
    "ridge_classifier.coef_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-0.31250723])"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# коэффициент свободного члена\n",
    "ridge_classifier.intercept_ "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### LogisticRegression\n",
    "Логистическая регрессия возвращает вероятности, с которыми объект принадлижит к тому или иному классу."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "log_regressor = linear_model.LogisticRegression(random_state = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "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=1, solver='liblinear', tol=0.0001,\n",
       "          verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "log_regressor.fit(train_data, train_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "lr_predictions = log_regressor.predict(test_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# построение вероятностей для предсказаений\n",
    "lr_proba_predictions = log_regressor.predict_proba(test_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0 0 0 1 0 0 0 0 0 1 0 1 0 1 1 1 0 0 0 1 0 0 0 1 1 1 1 0 0 1]\n"
     ]
    }
   ],
   "source": [
    "print test_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0 1 1 1 0 1 0 0 0 1 0 1 0 1 1 1 1 0 1 1 1 0 0 1 1 1 1 0 0 1]\n"
     ]
    }
   ],
   "source": [
    "print lr_predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 0.99873457  0.00126543]\n",
      " [ 0.38956803  0.61043197]\n",
      " [ 0.45000737  0.54999263]\n",
      " [ 0.00536816  0.99463184]\n",
      " [ 0.70965944  0.29034056]\n",
      " [ 0.23304944  0.76695056]\n",
      " [ 0.98983143  0.01016857]\n",
      " [ 0.9709671   0.0290329 ]\n",
      " [ 0.93909394  0.06090606]\n",
      " [ 0.15103335  0.84896665]\n",
      " [ 0.90324667  0.09675333]\n",
      " [ 0.0100448   0.9899552 ]\n",
      " [ 0.82338574  0.17661426]\n",
      " [ 0.05238427  0.94761573]\n",
      " [ 0.00443451  0.99556549]\n",
      " [ 0.00102911  0.99897089]\n",
      " [ 0.20817393  0.79182607]\n",
      " [ 0.91074564  0.08925436]\n",
      " [ 0.22580526  0.77419474]\n",
      " [ 0.00101709  0.99898291]\n",
      " [ 0.09600944  0.90399056]\n",
      " [ 0.99367624  0.00632376]\n",
      " [ 0.9779897   0.0220103 ]\n",
      " [ 0.29889823  0.70110177]\n",
      " [ 0.00210487  0.99789513]\n",
      " [ 0.00929423  0.99070577]\n",
      " [ 0.2670277   0.7329723 ]\n",
      " [ 0.85210662  0.14789338]\n",
      " [ 0.98871444  0.01128556]\n",
      " [ 0.00407572  0.99592428]]\n"
     ]
    }
   ],
   "source": [
    "print lr_proba_predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8\n"
     ]
    }
   ],
   "source": [
    "print metrics.accuracy_score(test_labels, lr_predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.866666666667\n"
     ]
    }
   ],
   "source": [
    "print metrics.accuracy_score(test_labels, ridge_predictions)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Оценка качества по cross-validation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### cross_val_score \n",
    "Оценка качества модели по кросс-валидации в соотвествии с заданными метрикой качества и стратегией кросс-валидации.\n",
    "Аргументы:\n",
    "Обьект модели классификатора\n",
    "Данные для классификации\n",
    "Метки данных\n",
    "Интересующая метрика\n",
    "cv = 10 - работает кросс-валидация по умолчанию, KFold - генерирует 10 фолдов (наборов)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "ridge_scoring = cross_validation.cross_val_score(ridge_classifier, blobs[0], blobs[1], scoring = 'accuracy', cv = 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "lr_scoring = cross_validation.cross_val_score(log_regressor, blobs[0], blobs[1], scoring = 'accuracy', cv = 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 0.7,  0.9,  0.9,  0.9,  1. ,  1. ,  0.7,  0.9,  0.9,  0.8])"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr_scoring"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ridge mean:0.88, max:1.0, min:0.7, std:0.0871779788708\n"
     ]
    }
   ],
   "source": [
    "print 'Ridge mean:{}, max:{}, min:{}, std:{}'.format(ridge_scoring.mean(), ridge_scoring.max(), \n",
    "                                                     ridge_scoring.min(), ridge_scoring.std())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Log mean:0.87, max:1.0, min:0.7, std:0.100498756211\n"
     ]
    }
   ],
   "source": [
    "print 'Log mean:{}, max:{}, min:{}, std:{}'.format(lr_scoring.mean(), lr_scoring.max(), \n",
    "                                                   lr_scoring.min(), lr_scoring.std())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### cross_val_score с заданными scorer и cv_strategy\n",
    "Спецификация кросс-валидации. Для нее необходимо создать метрику качества scorer, которая принимает 3 аргумента, модель, набор меток классов, набор предсказаний"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Передаем стандартную метрику metrics.accuracy_score\n",
    "scorer = metrics.make_scorer(metrics.accuracy_score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# специфицируем стратегию кросс-валидации применим StratifiedShuffleSplit\n",
    "# совершим 20 итераций\n",
    "cv_strategy = cross_validation.StratifiedShuffleSplit(blobs[1], n_iter = 20 , test_size = 0.3, random_state = 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "ridge_scoring = cross_validation.cross_val_score(ridge_classifier, blobs[0], blobs[1], scoring = scorer, cv = cv_strategy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "lr_scoring = cross_validation.cross_val_score(log_regressor, blobs[0], blobs[1], scoring = scorer, cv = cv_strategy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ridge mean:0.87, max:1.0, min:0.766666666667, std:0.0595352369983\n"
     ]
    }
   ],
   "source": [
    "print 'Ridge mean:{}, max:{}, min:{}, std:{}'.format(ridge_scoring.mean(), ridge_scoring.max(), \n",
    "                                                     ridge_scoring.min(), ridge_scoring.std())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Log mean:0.876666666667, max:1.0, min:0.766666666667, std:0.0615539510421\n"
     ]
    }
   ],
   "source": [
    "print 'Log mean:{}, max:{}, min:{}, std:{}'.format(lr_scoring.mean(), lr_scoring.max(), \n",
    "                                                   lr_scoring.min(), lr_scoring.std())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [Root]",
   "language": "python",
   "name": "Python [Root]"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}