Feature-selection-palooza.ipynb

{
 "metadata": {
  "name": "",
  "signature": "sha256:b35e26f8cd52676b9fbeedb0d43a4457a30e627a5383b546ad97b8b706a1d6f4"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Christopher:\n",
      "\n",
      "I wanted to follow up about the discussion for feature selection with a little more material.  I thought about it some more.\n",
      "\n",
      "If I recall correctly, we were talking about variable selection for a predictive churn model with a binary response.\n",
      "\n",
      "There are two approaches. \n",
      "\n",
      "1. Some classification methods return to us variable importance as a result of the method itself. We can just use the implied ranking and pick the best-k\n",
      "\n",
      "2. Any method which can predict also serves as a vehicle for a subset selection scheme. If the number of predictors is small, we can\n",
      "  * search the whole space of predictor combinations with our desired $k$. \n",
      "  * Othewise we can use stepwise regression, where we search trough remaining variables and add the one that most improves our accuracy. Rince, repeat, stop when you reach your desired $k$.\n",
      "\n",
      "[This paper](http://www.nesug.org/proceedings/nesug07/sa/sa07.pdf) from SAS which basically says use LASSO. However, Random Forest feature importance (explained below) has become more utilized, and the medhod I mentioned  [Boruta](https://m2.icm.edu.pl/boruta/) builds even on top of forest by trying out many many permutations in a kind of \"feature bootstrapping\". \n",
      "\n",
      "* note this is somewhat based on YHat's tutorial using the same data. They used k-means to cluster instead of LASSO."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import pandas as pd\n",
      "import numpy as np\n",
      "from sklearn import preprocessing, metrics, cross_validation, feature_selection, linear_model, svm,tree\n",
      "\n",
      "XOrig=pd.read_csv(\"https://raw.githubusercontent.com/EricChiang/churn/master/data/churn.csv\")\n",
      "# Some transfomatio and normalization \n",
      "XOrig.head()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "html": [
        "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n",
        "<table border=\"1\" class=\"dataframe\">\n",
        "  <thead>\n",
        "    <tr style=\"text-align: right;\">\n",
        "      <th></th>\n",
        "      <th>State</th>\n",
        "      <th>Account Length</th>\n",
        "      <th>Area Code</th>\n",
        "      <th>Phone</th>\n",
        "      <th>Int'l Plan</th>\n",
        "      <th>VMail Plan</th>\n",
        "      <th>VMail Message</th>\n",
        "      <th>Day Mins</th>\n",
        "      <th>Day Calls</th>\n",
        "      <th>Day Charge</th>\n",
        "      <th>...</th>\n",
        "      <th>Eve Calls</th>\n",
        "      <th>Eve Charge</th>\n",
        "      <th>Night Mins</th>\n",
        "      <th>Night Calls</th>\n",
        "      <th>Night Charge</th>\n",
        "      <th>Intl Mins</th>\n",
        "      <th>Intl Calls</th>\n",
        "      <th>Intl Charge</th>\n",
        "      <th>CustServ Calls</th>\n",
        "      <th>Churn?</th>\n",
        "    </tr>\n",
        "  </thead>\n",
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>0</th>\n",
        "      <td> KS</td>\n",
        "      <td> 128</td>\n",
        "      <td> 415</td>\n",
        "      <td> 382-4657</td>\n",
        "      <td>  no</td>\n",
        "      <td> yes</td>\n",
        "      <td> 25</td>\n",
        "      <td> 265.1</td>\n",
        "      <td> 110</td>\n",
        "      <td> 45.07</td>\n",
        "      <td>...</td>\n",
        "      <td>  99</td>\n",
        "      <td> 16.78</td>\n",
        "      <td> 244.7</td>\n",
        "      <td>  91</td>\n",
        "      <td> 11.01</td>\n",
        "      <td> 10.0</td>\n",
        "      <td> 3</td>\n",
        "      <td> 2.70</td>\n",
        "      <td> 1</td>\n",
        "      <td> False.</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>1</th>\n",
        "      <td> OH</td>\n",
        "      <td> 107</td>\n",
        "      <td> 415</td>\n",
        "      <td> 371-7191</td>\n",
        "      <td>  no</td>\n",
        "      <td> yes</td>\n",
        "      <td> 26</td>\n",
        "      <td> 161.6</td>\n",
        "      <td> 123</td>\n",
        "      <td> 27.47</td>\n",
        "      <td>...</td>\n",
        "      <td> 103</td>\n",
        "      <td> 16.62</td>\n",
        "      <td> 254.4</td>\n",
        "      <td> 103</td>\n",
        "      <td> 11.45</td>\n",
        "      <td> 13.7</td>\n",
        "      <td> 3</td>\n",
        "      <td> 3.70</td>\n",
        "      <td> 1</td>\n",
        "      <td> False.</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2</th>\n",
        "      <td> NJ</td>\n",
        "      <td> 137</td>\n",
        "      <td> 415</td>\n",
        "      <td> 358-1921</td>\n",
        "      <td>  no</td>\n",
        "      <td>  no</td>\n",
        "      <td>  0</td>\n",
        "      <td> 243.4</td>\n",
        "      <td> 114</td>\n",
        "      <td> 41.38</td>\n",
        "      <td>...</td>\n",
        "      <td> 110</td>\n",
        "      <td> 10.30</td>\n",
        "      <td> 162.6</td>\n",
        "      <td> 104</td>\n",
        "      <td>  7.32</td>\n",
        "      <td> 12.2</td>\n",
        "      <td> 5</td>\n",
        "      <td> 3.29</td>\n",
        "      <td> 0</td>\n",
        "      <td> False.</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>3</th>\n",
        "      <td> OH</td>\n",
        "      <td>  84</td>\n",
        "      <td> 408</td>\n",
        "      <td> 375-9999</td>\n",
        "      <td> yes</td>\n",
        "      <td>  no</td>\n",
        "      <td>  0</td>\n",
        "      <td> 299.4</td>\n",
        "      <td>  71</td>\n",
        "      <td> 50.90</td>\n",
        "      <td>...</td>\n",
        "      <td>  88</td>\n",
        "      <td>  5.26</td>\n",
        "      <td> 196.9</td>\n",
        "      <td>  89</td>\n",
        "      <td>  8.86</td>\n",
        "      <td>  6.6</td>\n",
        "      <td> 7</td>\n",
        "      <td> 1.78</td>\n",
        "      <td> 2</td>\n",
        "      <td> False.</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>4</th>\n",
        "      <td> OK</td>\n",
        "      <td>  75</td>\n",
        "      <td> 415</td>\n",
        "      <td> 330-6626</td>\n",
        "      <td> yes</td>\n",
        "      <td>  no</td>\n",
        "      <td>  0</td>\n",
        "      <td> 166.7</td>\n",
        "      <td> 113</td>\n",
        "      <td> 28.34</td>\n",
        "      <td>...</td>\n",
        "      <td> 122</td>\n",
        "      <td> 12.61</td>\n",
        "      <td> 186.9</td>\n",
        "      <td> 121</td>\n",
        "      <td>  8.41</td>\n",
        "      <td> 10.1</td>\n",
        "      <td> 3</td>\n",
        "      <td> 2.73</td>\n",
        "      <td> 3</td>\n",
        "      <td> False.</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
        "<p>5 rows \u00d7 21 columns</p>\n",
        "</div>"
       ],
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 282,
       "text": [
        "  State  Account Length  Area Code     Phone Int'l Plan VMail Plan  \\\n",
        "0    KS             128        415  382-4657         no        yes   \n",
        "1    OH             107        415  371-7191         no        yes   \n",
        "2    NJ             137        415  358-1921         no         no   \n",
        "3    OH              84        408  375-9999        yes         no   \n",
        "4    OK              75        415  330-6626        yes         no   \n",
        "\n",
        "   VMail Message  Day Mins  Day Calls  Day Charge   ...    Eve Calls  \\\n",
        "0             25     265.1        110       45.07   ...           99   \n",
        "1             26     161.6        123       27.47   ...          103   \n",
        "2              0     243.4        114       41.38   ...          110   \n",
        "3              0     299.4         71       50.90   ...           88   \n",
        "4              0     166.7        113       28.34   ...          122   \n",
        "\n",
        "   Eve Charge  Night Mins  Night Calls  Night Charge  Intl Mins  Intl Calls  \\\n",
        "0       16.78       244.7           91         11.01       10.0           3   \n",
        "1       16.62       254.4          103         11.45       13.7           3   \n",
        "2       10.30       162.6          104          7.32       12.2           5   \n",
        "3        5.26       196.9           89          8.86        6.6           7   \n",
        "4       12.61       186.9          121          8.41       10.1           3   \n",
        "\n",
        "   Intl Charge  CustServ Calls  Churn?  \n",
        "0         2.70               1  False.  \n",
        "1         3.70               1  False.  \n",
        "2         3.29               0  False.  \n",
        "3         1.78               2  False.  \n",
        "4         2.73               3  False.  \n",
        "\n",
        "[5 rows x 21 columns]"
       ]
      }
     ],
     "prompt_number": 282
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "Xch=XOrig\n",
      "Xch=Xch.drop(\"Area Code\",1)      # too specific\n",
      "Xch[\"Churn?\"]=np.where(Xch[\"Churn?\"]==\"True.\",1,0)\n",
      "y=Xch[\"Churn?\"]\n",
      "Xch[\"Int'l Plan\"]=np.where(Xch[\"Int'l Plan\"]=='yes',1,0)\n",
      "Xch[\"VMail Plan\"]=np.where(Xch[\"VMail Plan\"]=='yes',1,0)\n",
      "Xch=Xch.drop(\"State\",1)      # did not add anything to model and made it large\n",
      "Xch=Xch.drop(\"Phone\",1)      # too specific\n",
      "\n",
      "# scale and center\n",
      "X=Xch.as_matrix().astype(np.float)\n",
      "xform = preprocessing.StandardScaler()\n",
      "X=xform.fit_transform(X)\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 284
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Let's see if there are any obvious correlates that we should also drop."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import seaborn as sns\n",
      "sns.corrplot(X,annot=False,names=Xch.columns,diag_names=False)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 285,
       "text": [
        "<matplotlib.axes._subplots.AxesSubplot at 0x14c2edc10>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAcgAAAGNCAYAAACVNOE1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngyAIgmAAIZBBEARBMID/BwfZoRGxb2pFAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x15ed8d410>"
       ]
      }
     ],
     "prompt_number": 285
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "So we have some duplicate data, the mins and charges fields are identical information. Let's drop them. "
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "Xch=Xch.drop(\"Day Mins\",1)      # did not add anything to model and made it large\n",
      "Xch=Xch.drop(\"Eve Mins\",1)      # too specific\n",
      "Xch=Xch.drop(\"Night Mins\",1)      # too specific\n",
      "Xch=Xch.drop(\"Intl Mins\",1)      # too specific\n",
      "Xch=Xch.drop(\"VMail Message\",1)      # too specific\n",
      "# scale and center again\n",
      "X=Xch.as_matrix().astype(np.float)\n",
      "xform = preprocessing.StandardScaler()\n",
      "X=xform.fit_transform(X)\n",
      "\n",
      "sns.corrplot(X,annot=False,names=Xch.columns,diag_names=False)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 286,
       "text": [
        "<matplotlib.axes._subplots.AxesSubplot at 0x15346b590>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAcsAAAGNCAYAAAB+A1o2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhP8HjhO5K9fuOjIAAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x15f3ea510>"
       ]
      }
     ],
     "prompt_number": 286
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "That's better. Let's keep the correlation strength of the output var (Churn) to compare with other methods. And let's move it to Y."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "f_corr = np.array(Xch.corr()[\"Churn?\"].abs())[0:-1]   # drop last element\n",
      "X=X[:,0:-1]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 287
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "names=Xch.columns[0:-1]\n",
      "names"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 289,
       "text": [
        "Index(['Account Length', 'Int'l Plan', 'VMail Plan', 'Day Calls', 'Day Charge', 'Eve Calls', 'Eve Charge', 'Night Calls', 'Night Charge', 'Intl Calls', 'Intl Charge', 'CustServ Calls'], dtype='object')"
       ]
      }
     ],
     "prompt_number": 289
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "1. Logistic regression with L1 (LASSO)"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Ok, data prepped, Let's just throw logistic regression at it and look at the coefficients. Since we scaled all the data, the coefficient magnitude acutally means something. We can use the L1 penalization parameter \"C\" to force it into as sparse a solution as we want.\n"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "f1=linear_model.LogisticRegression(penalty='l1', C=0.01)\n",
      "f1.fit(X,y);\n",
      "f1_imp=abs(f1.coef_)[0]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 295
    },
    {
     "cell_type": "heading",
     "level": 3,
     "metadata": {},
     "source": [
      "2. Random Forest Importance Measure"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "We can use random forest to compute importances. It works like this, roughly.\n",
      "During the algo, random forest creates a series of small decision trees, each only using subset of predictors.\n",
      "\n",
      "The algo holds out some data from the formation of each tree. The raemainder is classified through the tree and an accuracy score is computed. Here's the clever bit. For each predictor that made it into a given tree, the accuracy score is recomputed **with that predictor's values ramdomly permuted**. The original score is subtracted from the permuted data version. If the feature was indeed informative it should on average add accurace.\n",
      "\n",
      "Then we sum up, over all trees, how much cumulative improvement each variable did. That's our feature importance."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "f2=tree.ExtraTreeClassifier()\n",
      "f2.fit(X,y)\n",
      "f2_imp=f2.feature_importances_"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 296
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Let's take the top 6, rank them and compare."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import matplotlib.pyplot as plt\n",
      "\n",
      "# let's rank the top 6 attributes of each.\n",
      "f_ncorr=f_corr.argsort()-5\n",
      "f_ncorr[f_ncorr<1]=0\n",
      "\n",
      "f1_nimp=f1_imp.argsort()-5\n",
      "f1_nimp[f1_nimp<1]=0\n",
      "f2_nimp=f2_imp.argsort()-5\n",
      "f2_nimp[f2_nimp<1]=0\n",
      "\n",
      "plt.figure(figsize=(10,10), dpi=80)\n",
      "\n",
      "idx=np.arange(len(f1_imp))*2.5\n",
      "plt.barh(idx+0.6,list(f_ncorr),height=0.5,color='r', label=\"Correlation with y\")\n",
      "plt.barh(idx-0.0,list(f1_nimp),height=0.5,color='g', label=\"L1 Logit\")\n",
      "plt.barh(idx-0.6,list(f2_nimp),height=0.5,color='b', label=\"Rand Forest\")\n",
      "plt.legend(loc='upper right')\n",
      "plt.yticks(idx,list(names));"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAooAAAJMCAYAAAB9xYclAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrs10TOD2AAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x1533f3150>"
       ]
      }
     ],
     "prompt_number": 362
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "What conclusins can we draw from this ? First, people look cost sensitive more than loyal. Account length is nowhere near informative as usage counts. Finally, those who make international calls are far more likely to switch. Perhaps the costs are so much higher compared to domenstic long distance rates that they can save significant money by switching."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "OK. Time to shake our money maker. Let's predict churn. I'm going to only use the top 6 predictors for each. First let's see if we have really unbalanced response classes. I should probably have checked this earlier."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from scipy.stats import itemfreq\n",
      "itemfreq(y)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 324,
       "text": [
        "array([[   0, 2850],\n",
        "       [   1,  483]])"
       ]
      }
     ],
     "prompt_number": 324
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Yeah a bit unbalanced. Given that we'll use stratified 10 fold cross validation."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "X_lasso=X[:,f1_nimp>0]   # only use what lasso wanted\n",
      "X_rf   =X[:,f2_nimp>0]   # only use what rf wanted\n",
      "y_rf=np.zeros(y.shape)\n",
      "y_lasso=np.zeros(y.shape)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 363
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from sklearn import cross_validation,metrics\n",
      "cv=cross_validation.StratifiedKFold(y,n_folds=10)\n",
      "\n",
      "for X_train,X_test in cv:\n",
      "    f_lasso=linear_model.LogisticRegression(penalty='l1',C=0.1)\n",
      "    f_lasso.fit(X_lasso[X_train],y[X_train])\n",
      "    y_lasso[X_test]=f_lasso.predict(X_lasso[X_test])\n",
      "    \n",
      "    f_rf   =tree.ExtraTreeClassifier()\n",
      "    f_rf.fit(X_rf[X_train],y[X_train])\n",
      "    y_rf[X_test]=f_rf.predict(X_rf[X_test])\n",
      "    \n",
      "    \n",
      "print(\"LASSO accuracy with 6 vars: \" + str(metrics.accuracy_score(y,y_lasso)))\n",
      "print(metrics.confusion_matrix(y,y_lasso)/y.shape)\n",
      "\n",
      "print(\"Random Forest accuracy with 6 vars: \" + str(metrics.accuracy_score(y,y_rf)))\n",
      "print(metrics.confusion_matrix(y,y_rf)/y.shape)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "LASSO accuracy with 6 vars: 0.85598559856\n",
        "[[ 0.85508551  0.        ]\n",
        " [ 0.1440144   0.00090009]]\n",
        "Random Forest accuracy with 6 vars: 0.791479147915\n",
        "[[ 0.74617462  0.10891089]\n",
        " [ 0.09960996  0.04530453]]\n"
       ]
      }
     ],
     "prompt_number": 365
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "LASSO did much better with limited variables. Generally random forest does better in most cases, but it's dataset dependent."
     ]
    }
   ],
   "metadata": {}
  }
 ]
}