13_feat-sele/13_feat-sele_code.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "STAT 479: Machine Learning (Fall 2018)  \n",
    "Instructor: Sebastian Raschka (sraschka@wisc.edu)  \n",
    "Course website: http://pages.stat.wisc.edu/~sraschka/teaching/stat479-fs2018/"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# L13: Dimensionality Reduction 1 -- Feature Selection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sebastian Raschka 2018-11-26 \n",
      "\n",
      "CPython 3.6.5\n",
      "IPython 6.5.0\n",
      "\n",
      "sklearn 0.20.0\n",
      "mlxtend 0.15.0dev\n"
     ]
    }
   ],
   "source": [
    "%load_ext watermark\n",
    "%watermark -a 'Sebastian Raschka' -d -p sklearn,mlxtend -v"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Iris Overview"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x720 with 16 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "from mlxtend.data import iris_data\n",
    "from mlxtend.plotting import scatterplotmatrix\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "X, y = iris_data()\n",
    "\n",
    "\n",
    "names = ['sepal length [cm]', 'sepal width [cm]',\n",
    "         'petal length [cm]', 'petal width [cm]']\n",
    "\n",
    "fig, axes = scatterplotmatrix(X[y==0], figsize=(12, 10), alpha=0.5)\n",
    "fig, axes = scatterplotmatrix(X[y==1], fig_axes=(fig, axes), alpha=0.5)\n",
    "fig, axes = scatterplotmatrix(X[y==2], fig_axes=(fig, axes), alpha=0.5, names=names)\n",
    "\n",
    "plt.legend(['setosa', 'versicolor', 'virginica'])\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Suppose in a real-world, high dimensional dataset, we cannot intuitively tell which the good features are"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Feature Importance from Linear Model Coefficients"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Binary Class"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![](code-figures/logreg.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy 96.67%\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    X[50:150], y[50:150], test_size=0.3, random_state=1, stratify=y[50:150])\n",
    "\n",
    "\n",
    "pipe = make_pipeline(StandardScaler(),\n",
    "                     # params below actually not necessary:\n",
    "                     LogisticRegression(solver='newton-cg', multi_class='multinomial'))\n",
    "\n",
    "pipe.fit(X_train, y_train)\n",
    "print('Test Accuracy %.2f%%' % (100*pipe.score(X_test, y_test)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.09211828, -0.29927205,  1.27812501,  1.22789905]])"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr_model = pipe.named_steps.logisticregression\n",
    "lr_model.coef_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.09211828, 0.29927205, 1.27812501, 1.22789905]])"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.abs(lr_model.coef_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "\n",
    "ax.bar(range(X_train.shape[1]), np.abs(lr_model.coef_.flatten()))\n",
    "ax.set_xlim([-1, X.shape[1]])\n",
    "ax.set_xticks(range(X_train.shape[1]))\n",
    "ax.set_xticklabels(['sepal length', 'sepal width', 'petal length', 'petal width'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Multi-Class"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![](code-figures/multinomial-logreg.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy 97.78%\n"
     ]
    }
   ],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    X, y, test_size=0.3, random_state=1, stratify=y)\n",
    "\n",
    "\n",
    "pipe = make_pipeline(StandardScaler(),\n",
    "                     LogisticRegression(solver='newton-cg', multi_class='multinomial'))\n",
    "\n",
    "pipe.fit(X_train, y_train)\n",
    "print('Test Accuracy %.2f%%' % (100*pipe.score(X_test, y_test)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-1.05554231,  1.01406239, -1.75805264, -1.60180046],\n",
       "       [ 0.22718366, -0.26599486, -0.13917565, -0.58186585],\n",
       "       [ 0.82835866, -0.74806753,  1.89722829,  2.18366631]])"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr_model = pipe.named_steps.logisticregression\n",
    "lr_model.coef_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([2.11108463, 2.02812478, 3.79445658, 4.36733262])"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "coefs = np.sum(np.abs(lr_model.coef_), axis=0)\n",
    "coefs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "\n",
    "ax.bar(range(X_train.shape[1]), coefs)\n",
    "ax.set_xlim([-1, X.shape[1]])\n",
    "ax.set_xticks(range(X_train.shape[1]))\n",
    "ax.set_xticklabels(['sepal length', 'sepal width', 'petal length', 'petal width'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Feature Importance via Permutation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- http://rasbt.github.io/mlxtend/user_guide/evaluate/feature_importance_permutation/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "from mlxtend.evaluate import feature_importance_permutation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.neighbors import KNeighborsClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy 95.56%\n"
     ]
    }
   ],
   "source": [
    "pipe = make_pipeline(StandardScaler(),\n",
    "                     KNeighborsClassifier(n_neighbors=3))\n",
    "\n",
    "pipe.fit(X_train, y_train)\n",
    "print('Test Accuracy %.2f%%' % (100*pipe.score(X_test, y_test)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.03711111, 0.07066667, 0.20488889, 0.31022222])"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "imp_vals, _ = feature_importance_permutation(\n",
    "    predict_method=pipe.predict, \n",
    "    X=X_test,\n",
    "    y=y_test,\n",
    "    metric='accuracy',\n",
    "    num_rounds=100,\n",
    "    seed=1)\n",
    "\n",
    "imp_vals"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "\n",
    "ax.bar(range(X_train.shape[1]), imp_vals)\n",
    "ax.set_xlim([-1, X.shape[1]])\n",
    "ax.set_xticks(range(X_train.shape[1]))\n",
    "ax.set_xticklabels(['sepal length', 'sepal width', 'petal length', 'petal width'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Random Forest Feature Importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "\n",
    "forest = RandomForestClassifier(n_estimators=100, random_state=123)\n",
    "forest.fit(X_train, y_train);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "\n",
    "ax.bar(range(X_train.shape[1]), forest.feature_importances_)\n",
    "ax.set_xlim([-1, X.shape[1]])\n",
    "ax.set_xticks(range(X_train.shape[1]))\n",
    "ax.set_xticklabels(['sepal length', 'sepal width', 'petal length', 'petal width'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Sequential Feature Selection"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- http://rasbt.github.io/mlxtend/user_guide/feature_selection/SequentialFeatureSelector/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.\n",
      "[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    0.1s finished\n",
      "\n",
      "[2018-11-26 22:23:11] Features: 1/4 -- score: 0.9416666666666667[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.\n",
      "[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    0.1s finished\n",
      "\n",
      "[2018-11-26 22:23:11] Features: 2/4 -- score: 0.9583333333333334[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.\n",
      "[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    0.0s finished\n",
      "\n",
      "[2018-11-26 22:23:11] Features: 3/4 -- score: 0.9333333333333333[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.\n",
      "[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.0s finished\n",
      "\n",
      "[2018-11-26 22:23:11] Features: 4/4 -- score: 0.9333333333333333"
     ]
    }
   ],
   "source": [
    "from mlxtend.feature_selection import SequentialFeatureSelector as SFS\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "\n",
    "\n",
    "pipe = make_pipeline(StandardScaler(),\n",
    "                     KNeighborsClassifier(n_neighbors=5))\n",
    "\n",
    "sfs1 = SFS(pipe, \n",
    "           k_features=4, \n",
    "           forward=True, \n",
    "           floating=False, \n",
    "           verbose=2,\n",
    "           scoring='accuracy',\n",
    "           cv=10)\n",
    "\n",
    "sfs1 = sfs1.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{1: {'feature_idx': (2,),\n",
       "  'cv_scores': array([1.        , 0.91666667, 1.        , 0.91666667, 0.91666667,\n",
       "         1.        , 1.        , 0.88888889, 0.88888889, 0.88888889]),\n",
       "  'avg_score': 0.9416666666666667,\n",
       "  'feature_names': ('2',)},\n",
       " 2: {'feature_idx': (2, 3),\n",
       "  'cv_scores': array([1.        , 1.        , 1.        , 1.        , 0.91666667,\n",
       "         1.        , 0.88888889, 1.        , 0.88888889, 0.88888889]),\n",
       "  'avg_score': 0.9583333333333334,\n",
       "  'feature_names': ('2', '3')},\n",
       " 3: {'feature_idx': (0, 2, 3),\n",
       "  'cv_scores': array([1.        , 0.91666667, 0.91666667, 0.91666667, 0.91666667,\n",
       "         1.        , 0.88888889, 1.        , 0.88888889, 0.88888889]),\n",
       "  'avg_score': 0.9333333333333333,\n",
       "  'feature_names': ('0', '2', '3')},\n",
       " 4: {'feature_idx': (0, 1, 2, 3),\n",
       "  'cv_scores': array([1.        , 1.        , 0.91666667, 0.83333333, 0.91666667,\n",
       "         1.        , 1.        , 0.88888889, 0.88888889, 0.88888889]),\n",
       "  'avg_score': 0.9333333333333333,\n",
       "  'feature_names': ('0', '1', '2', '3')}}"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sfs1.subsets_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>avg_score</th>\n",
       "      <th>ci_bound</th>\n",
       "      <th>cv_scores</th>\n",
       "      <th>feature_idx</th>\n",
       "      <th>feature_names</th>\n",
       "      <th>std_dev</th>\n",
       "      <th>std_err</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.941667</td>\n",
       "      <td>0.0295001</td>\n",
       "      <td>[1.0, 0.9166666666666666, 1.0, 0.9166666666666...</td>\n",
       "      <td>(2,)</td>\n",
       "      <td>(2,)</td>\n",
       "      <td>0.0488289</td>\n",
       "      <td>0.0162763</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.958333</td>\n",
       "      <td>0.0311713</td>\n",
       "      <td>[1.0, 1.0, 1.0, 1.0, 0.9166666666666666, 1.0, ...</td>\n",
       "      <td>(2, 3)</td>\n",
       "      <td>(2, 3)</td>\n",
       "      <td>0.0515949</td>\n",
       "      <td>0.0171983</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.933333</td>\n",
       "      <td>0.0272676</td>\n",
       "      <td>[1.0, 0.9166666666666666, 0.9166666666666666, ...</td>\n",
       "      <td>(0, 2, 3)</td>\n",
       "      <td>(0, 2, 3)</td>\n",
       "      <td>0.0451335</td>\n",
       "      <td>0.0150445</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.933333</td>\n",
       "      <td>0.035362</td>\n",
       "      <td>[1.0, 1.0, 0.9166666666666666, 0.8333333333333...</td>\n",
       "      <td>(0, 1, 2, 3)</td>\n",
       "      <td>(0, 1, 2, 3)</td>\n",
       "      <td>0.0585314</td>\n",
       "      <td>0.0195105</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  avg_score   ci_bound                                          cv_scores  \\\n",
       "1  0.941667  0.0295001  [1.0, 0.9166666666666666, 1.0, 0.9166666666666...   \n",
       "2  0.958333  0.0311713  [1.0, 1.0, 1.0, 1.0, 0.9166666666666666, 1.0, ...   \n",
       "3  0.933333  0.0272676  [1.0, 0.9166666666666666, 0.9166666666666666, ...   \n",
       "4  0.933333   0.035362  [1.0, 1.0, 0.9166666666666666, 0.8333333333333...   \n",
       "\n",
       "    feature_idx feature_names    std_dev    std_err  \n",
       "1          (2,)          (2,)  0.0488289  0.0162763  \n",
       "2        (2, 3)        (2, 3)  0.0515949  0.0171983  \n",
       "3     (0, 2, 3)     (0, 2, 3)  0.0451335  0.0150445  \n",
       "4  (0, 1, 2, 3)  (0, 1, 2, 3)  0.0585314  0.0195105  "
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "\n",
    "pd.DataFrame.from_dict(sfs1.get_metric_dict(confidence_interval=0.90)).T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from mlxtend.plotting import plot_sequential_feature_selection as plot_sfs\n",
    "\n",
    "\n",
    "fig1 = plot_sfs(sfs1.get_metric_dict(), kind='std_dev')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Pipeline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy 100.00%\n"
     ]
    }
   ],
   "source": [
    "from mlxtend.feature_selection import SequentialFeatureSelector as SFS\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "\n",
    "\n",
    "pipe = make_pipeline(StandardScaler(),\n",
    "                     SFS(estimator=KNeighborsClassifier(n_neighbors=5), k_features=2),\n",
    "                     KNeighborsClassifier(n_neighbors=5))\n",
    "\n",
    "\n",
    "pipe.fit(X_train, y_train)\n",
    "print('Test Accuracy %.2f%%' % (100*pipe.score(X_test, y_test)))"
   ]
  }
 ],
 "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": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}