sklearn/Model Selection (cv, grid search).ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Top\n",
    "http://scikit-learn.org/stable/tutorial/statistical_inference/model_selection.html\n",
    "\n",
    "## Relevant functions\n",
    "Modules\n",
    "- [cross_validation](http://scikit-learn.org/stable/modules/classes.html#module-sklearn.cross_validation)\n",
    "  - [`KFold(n, n_folds=3, indices=None, shuffle=False, random_state=None)[source]`](http://scikit-learn.org/stable/modules/generated/sklearn.cross_validation.KFold.html)\n",
    "  - [`StratifiedKFold(y, n_folds=3, indices=None, shuffle=False, random_state=None)`](http://scikit-learn.org/stable/modules/generated/sklearn.cross_validation.StratifiedKFold.html)\n",
    "- [grid_search](http://scikit-learn.org/stable/modules/classes.html#module-sklearn.grid_search)\n",
    "    - [`grid_search.GridSearchCV(estimator, param_grid)`](http://scikit-learn.org/stable/modules/generated/sklearn.grid_search.GridSearchCV.html)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.97999999999999998"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn import datasets, svm\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "digits = datasets.load_digits()\n",
    "X_digits = digits.data\n",
    "y_digits = digits.target\n",
    "svc = svm.SVC(C=1, kernel='linear')\n",
    "svc.fit(X_digits[:-100], y_digits[:-100]).score(X_digits[-100:], y_digits[-100:])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Cross validation (for obtaining performance measure)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Approach1: np.array_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Scores in each fold (3-CV):  [0.93489148580968284, 0.95659432387312182, 0.93989983305509184]\n"
     ]
    }
   ],
   "source": [
    "X_folds = np.array_split(X_digits, 3)\n",
    "y_folds = np.array_split(y_digits, 3)\n",
    "scores = list()\n",
    "for k in range(3):\n",
    "    # We use 'list' to copy, in order to 'pop' later on\n",
    "    X_train = list(X_folds)\n",
    "    X_test  = X_train.pop(k)\n",
    "    X_train = np.concatenate(X_train)\n",
    "    y_train = list(y_folds)\n",
    "    y_test  = y_train.pop(k)\n",
    "    y_train = np.concatenate(y_train)\n",
    "    scores.append(svc.fit(X_train, y_train).score(X_test, y_test))\n",
    "print \"Scores in each fold (3-CV): \", scores"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Approach2: use cv generators"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train: [2 3 4 5] | test: [0 1]\n",
      "Train: [0 1 4 5] | test: [2 3]\n",
      "Train: [0 1 2 3] | test: [4 5]\n",
      "Scores in each fold (3-CV):  [0.93489148580968284, 0.95659432387312182, 0.93989983305509184]\n"
     ]
    }
   ],
   "source": [
    "from sklearn import cross_validation\n",
    "k_fold = cross_validation.KFold(n=6, n_folds=3)\n",
    "for train_indices, test_indices in k_fold:\n",
    "     print('Train: %s | test: %s' % (train_indices, test_indices))\n",
    "\n",
    "# implement CV here\n",
    "kfold = cross_validation.KFold(len(X_digits), n_folds=3)\n",
    "\n",
    "# wonky but list comprehension is possible (as cv instance is iterable)\n",
    "[svc.fit(X_digits[train], y_digits[train]).score(X_digits[test], y_digits[test])\n",
    "         for train, test in kfold]\n",
    "print \"Scores in each fold (3-CV): \", scores"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Grid-search and cv\n",
    "http://scikit-learn.org/stable/modules/generated/sklearn.grid_search.GridSearchCV.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-13.81551056, -12.53629662, -11.25708268,  -9.97786874,\n",
       "        -8.6986548 ,  -7.41944086,  -6.14022691,  -4.86101297,\n",
       "        -3.58179903,  -2.30258509])"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAASIAAADOCAYAAACaTw/7AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIYqjICKEKO7/AYV/6bUnkElIAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f84f1531390>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.grid_search import GridSearchCV\n",
    "Cs = np.logspace(-6, -1,10)\n",
    "plt.figure(figsize=(3,2))\n",
    "plt.plot(Cs, label='meh' )\n",
    "plt.plot(np.log(Cs), 'r', label='log(Cs)')\n",
    "plt.legend(loc='best')\n",
    "np.log(Cs) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Notes**\n",
    " - gridsearch instantation accepts estimator, dict of gridsearch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.925\n",
      "0.00774263682681\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.94353826850690092"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# create an instance of GridSearch object\n",
    "clf = GridSearchCV(estimator=svc, param_grid=dict(C=Cs),n_jobs=-1)\n",
    "Xtr,Xts=X_digits[:1000],X_digits[1000:]\n",
    "ytr,yts=y_digits[:1000],y_digits[1000:]\n",
    "\n",
    "# tune (notice \"fit\" method can be applied)\n",
    "clf.fit(Xtr,ytr)        \n",
    "\n",
    "print clf.best_score_                                  \n",
    "print clf.best_estimator_.C                            \n",
    "\n",
    "# Prediction performance on test set is not as good as on train set\n",
    "clf.score(Xts,yts)      "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Explore what's inside GridSearCV instance\n",
    "\n",
    "Some notes about pandas:\n",
    "- df has no `str` method\n",
    "- Series has no `query` method"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>best_estimator_</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>best_params_</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>best_score_</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>grid_scores_</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>scorer_</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  0\n",
       "25  best_estimator_\n",
       "26     best_params_\n",
       "27      best_score_\n",
       "35     grid_scores_\n",
       "46          scorer_"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# some notes about pandas:\n",
    "# - df has no \"str\" method\n",
    "# - Series has no \"query\" method\n",
    "\n",
    "# pd.set_option('display.max_rows', 5)\n",
    "# pd.set_option('precision',3)\n",
    "# pd.DataFrame(dir(clf)).ix[25:].str.upper\n",
    "\n",
    "tmp = pd.Series(dir(clf))[25:] # <- Use Series since DF has no \"str\" method\n",
    "df = pd.DataFrame(tmp[tmp.str.endswith('_')]) # <- convert to DataFrame for display purpose\n",
    "df\n",
    "# tmp.quantile\n",
    "# tmp.str.endswith('_') # <- \"str\" in pandas only available for Series (me thinks...)\n",
    "\n",
    "# tmp[tmp[0].str.('_')]\n",
    "# tmp."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "clf.best_estimator_ = SVC(C=0.0077426368268112772, cache_size=200, class_weight=None, coef0=0.0,\n",
      "  degree=3, gamma=0.0, kernel='linear', max_iter=-1, probability=False,\n",
      "  random_state=None, shrinking=True, tol=0.001, verbose=False)\n",
      "\n",
      "clf.best_params_ = {'C': 0.0077426368268112772}\n",
      "\n",
      "clf.best_score_ = 0.925\n",
      "\n",
      "clf.grid_scores_ = [mean: 0.10400, std: 0.00073, params: {'C': 9.9999999999999995e-07}, mean: 0.10400, std: 0.00073, params: {'C': 3.5938136638046257e-06}, mean: 0.50600, std: 0.00489, params: {'C': 1.2915496650148827e-05}, mean: 0.84600, std: 0.00739, params: {'C': 4.6415888336127818e-05}, mean: 0.90500, std: 0.01511, params: {'C': 0.00016681005372000591}, mean: 0.92000, std: 0.01047, params: {'C': 0.00059948425031894088}, mean: 0.92100, std: 0.02001, params: {'C': 0.0021544346900318843}, mean: 0.92500, std: 0.01871, params: {'C': 0.0077426368268112772}, mean: 0.92400, std: 0.01729, params: {'C': 0.02782559402207126}, mean: 0.92400, std: 0.01729, params: {'C': 0.10000000000000001}]\n",
      "\n",
      "clf.scorer_ = <function _passthrough_scorer at 0x7f84f853db90>\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[mean: 0.10400, std: 0.00073, params: {'C': 9.9999999999999995e-07},\n",
       " mean: 0.10400, std: 0.00073, params: {'C': 3.5938136638046257e-06},\n",
       " mean: 0.50600, std: 0.00489, params: {'C': 1.2915496650148827e-05},\n",
       " mean: 0.84600, std: 0.00739, params: {'C': 4.6415888336127818e-05},\n",
       " mean: 0.90500, std: 0.01511, params: {'C': 0.00016681005372000591},\n",
       " mean: 0.92000, std: 0.01047, params: {'C': 0.00059948425031894088},\n",
       " mean: 0.92100, std: 0.02001, params: {'C': 0.0021544346900318843},\n",
       " mean: 0.92500, std: 0.01871, params: {'C': 0.0077426368268112772},\n",
       " mean: 0.92400, std: 0.01729, params: {'C': 0.02782559402207126},\n",
       " mean: 0.92400, std: 0.01729, params: {'C': 0.10000000000000001}]"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "for k in df.values:\n",
    "    # print attributes ending with underscore\n",
    "#     print k[0]\n",
    "    print \"\\nclf.{} = {}\".format(k[0], getattr(clf,k[0]))\n",
    "\n",
    "# print \"\\nclf.best_estimator_ = \",clf.best_estimator_\n",
    "# print \"\\nclf.best_params_\",clf.best_params_\n",
    "# print \"\\nclf.best_score_\",clf.best_score_\n",
    "# print \"\\nclf.scorer_\",clf.scorer_\n",
    "clf.grid_scores_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>C</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>{u'C': 1e-06}</td>\n",
       "      <td>0.104</td>\n",
       "      <td>[0.104166666667, 0.104790419162, 0.10303030303]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>{u'C': 3.5938136638e-06}</td>\n",
       "      <td>0.104</td>\n",
       "      <td>[0.104166666667, 0.104790419162, 0.10303030303]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>{u'C': 1.29154966501e-05}</td>\n",
       "      <td>0.506</td>\n",
       "      <td>[0.5, 0.511976047904, 0.506060606061]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>{u'C': 4.64158883361e-05}</td>\n",
       "      <td>0.846</td>\n",
       "      <td>[0.839285714286, 0.85628742515, 0.842424242424]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>{u'C': 0.00016681005372}</td>\n",
       "      <td>0.905</td>\n",
       "      <td>[0.883928571429, 0.913173652695, 0.918181818182]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>{u'C': 0.000599484250319}</td>\n",
       "      <td>0.920</td>\n",
       "      <td>[0.907738095238, 0.919161676647, 0.933333333333]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>{u'C': 0.00215443469003}</td>\n",
       "      <td>0.921</td>\n",
       "      <td>[0.91369047619, 0.90119760479, 0.948484848485]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>{u'C': 0.00774263682681}</td>\n",
       "      <td>0.925</td>\n",
       "      <td>[0.91369047619, 0.910179640719, 0.951515151515]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>{u'C': 0.0278255940221}</td>\n",
       "      <td>0.924</td>\n",
       "      <td>[0.91369047619, 0.910179640719, 0.948484848485]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>{u'C': 0.1}</td>\n",
       "      <td>0.924</td>\n",
       "      <td>[0.91369047619, 0.910179640719, 0.948484848485]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 73,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.DataFrame(clf.grid_scores_,columns=['C','mean','std'])\n",
    "\n",
    "# from IPython.display import HTML\n",
    "# h = HTML(df.to_html());"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'best_estimator_': SVC(C=0.0077426368268112772, cache_size=200, class_weight=None, coef0=0.0,\n",
       "   degree=3, gamma=0.0, kernel='linear', max_iter=-1, probability=False,\n",
       "   random_state=None, shrinking=True, tol=0.001, verbose=False),\n",
       " 'best_params_': {'C': 0.0077426368268112772},\n",
       " 'best_score_': 0.92500000000000004,\n",
       " 'cv': None,\n",
       " 'error_score': 'raise',\n",
       " 'estimator': SVC(C=1, cache_size=200, class_weight=None, coef0=0.0, degree=3, gamma=0.0,\n",
       "   kernel='linear', max_iter=-1, probability=False, random_state=None,\n",
       "   shrinking=True, tol=0.001, verbose=False),\n",
       " 'fit_params': {},\n",
       " 'grid_scores_': [mean: 0.10400, std: 0.00073, params: {'C': 9.9999999999999995e-07},\n",
       "  mean: 0.10400, std: 0.00073, params: {'C': 3.5938136638046257e-06},\n",
       "  mean: 0.50600, std: 0.00489, params: {'C': 1.2915496650148827e-05},\n",
       "  mean: 0.84600, std: 0.00739, params: {'C': 4.6415888336127818e-05},\n",
       "  mean: 0.90500, std: 0.01511, params: {'C': 0.00016681005372000591},\n",
       "  mean: 0.92000, std: 0.01047, params: {'C': 0.00059948425031894088},\n",
       "  mean: 0.92100, std: 0.02001, params: {'C': 0.0021544346900318843},\n",
       "  mean: 0.92500, std: 0.01871, params: {'C': 0.0077426368268112772},\n",
       "  mean: 0.92400, std: 0.01729, params: {'C': 0.02782559402207126},\n",
       "  mean: 0.92400, std: 0.01729, params: {'C': 0.10000000000000001}],\n",
       " 'iid': True,\n",
       " 'n_jobs': -1,\n",
       " 'param_grid': {'C': array([  1.00000000e-06,   3.59381366e-06,   1.29154967e-05,\n",
       "           4.64158883e-05,   1.66810054e-04,   5.99484250e-04,\n",
       "           2.15443469e-03,   7.74263683e-03,   2.78255940e-02,\n",
       "           1.00000000e-01])},\n",
       " 'pre_dispatch': '2*n_jobs',\n",
       " 'refit': True,\n",
       " 'scorer_': <function sklearn.metrics.scorer._passthrough_scorer>,\n",
       " 'scoring': None,\n",
       " 'verbose': 0}"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# fuck...this probably is better\n",
    "clf.__dict__"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Weird....GridSearchCV doesn't seem to return CV-indices...."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Bunch of examples"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## RFE with CV (GridSearch class not used)\n",
    "- ([link](http://scikit-learn.org/stable/auto_examples/feature_selection/plot_rfe_with_cross_validation.html#example-feature-selection-plot-rfe-with-cross-validation-py))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimal number of features : 3\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiMAAAGHCAYAAABiT1LUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nbz4slywAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f852e0730d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.svm import SVC\n",
    "from sklearn.cross_validation import StratifiedKFold\n",
    "from sklearn.feature_selection import RFECV\n",
    "from sklearn.datasets import make_classification\n",
    "\n",
    "# Build a classification task using 3 informative features\n",
    "X, y = make_classification(n_samples=1000, n_features=25, n_informative=3,\n",
    "                           n_redundant=2, n_repeated=0, n_classes=8,\n",
    "                           n_clusters_per_class=1, random_state=0)\n",
    "\n",
    "svc = SVC(kernel=\"linear\")\n",
    "\n",
    "# Create the RFE object and compute a cross-validated score.\n",
    "# The \"accuracy\" scoring is proportional to the number of correct classifications\n",
    "rfecv = RFECV(estimator=svc, step=1, cv=StratifiedKFold(y, 2),scoring='accuracy') \n",
    "rfecv.fit(X, y)\n",
    "\n",
    "print(\"Optimal number of features : %d\" % rfecv.n_features_)\n",
    "\n",
    "# Plot number of features VS. cross-validation scores\n",
    "plt.figure()\n",
    "plt.xlabel(\"Number of features selected\")\n",
    "plt.ylabel(\"CV score (nb of correct classifications)\")\n",
    "plt.plot(range(1, len(rfecv.grid_scores_) + 1), rfecv.grid_scores_)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['__abstractmethods__',\n",
       " '__class__',\n",
       " '__delattr__',\n",
       " '__dict__',\n",
       " '__doc__',\n",
       " '__format__',\n",
       " '__getattribute__',\n",
       " '__hash__',\n",
       " '__init__',\n",
       " '__module__',\n",
       " '__new__',\n",
       " '__reduce__',\n",
       " '__reduce_ex__',\n",
       " '__repr__',\n",
       " '__setattr__',\n",
       " '__sizeof__',\n",
       " '__str__',\n",
       " '__subclasshook__',\n",
       " '__weakref__',\n",
       " '_abc_cache',\n",
       " '_abc_negative_cache',\n",
       " '_abc_negative_cache_version',\n",
       " '_abc_registry',\n",
       " '_get_param_names',\n",
       " '_get_support_mask',\n",
       " 'cv',\n",
       " 'decision_function',\n",
       " 'estimator',\n",
       " 'estimator_',\n",
       " 'estimator_params',\n",
       " 'fit',\n",
       " 'fit_transform',\n",
       " 'get_params',\n",
       " 'get_support',\n",
       " 'grid_scores_',\n",
       " 'inverse_transform',\n",
       " 'n_features_',\n",
       " 'predict',\n",
       " 'predict_log_proba',\n",
       " 'predict_proba',\n",
       " 'ranking_',\n",
       " 'score',\n",
       " 'scoring',\n",
       " 'set_params',\n",
       " 'step',\n",
       " 'support_',\n",
       " 'transform',\n",
       " 'verbose']"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dir(rfecv)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}