{ "metadata": { "name": "", "signature": "sha256:6b0ab938c33123043d28a9f4abc0ffb565e858da304658e2c1afbf3d8ee371a0" }, "nbformat": 3, "nbformat_minor": 0, "worksheets": [ { "cells": [ { "cell_type": "code", "collapsed": false, "input": [ "%matplotlib inline\n", "import matplotlib.pyplot as plt\n", "\n", "# Some nice default configuration for plots\n", "plt.rcParams['figure.figsize'] = 10, 7.5\n", "plt.rcParams['axes.grid'] = True\n", "plt.gray()\n", "\n", "import numpy as np\n", "import pandas as pd" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "display_data", "text": [ "" ] } ], "prompt_number": 1 }, { "cell_type": "heading", "level": 1, "metadata": {}, "source": [ "11. Measuring performance in Classification Models" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We now turn our focus to building and evaluating models for a categorical response. Although many of the regression modeling techniques can also be used for classification, the way we evaluate model performance is necessarily very different since metrics like RMSE and $R^2$ are not appropriate in the context of classification. We will take an in-depth look at the different aspects of classification model predictions and how these relate to the question of interest." ] }, { "cell_type": "heading", "level": 2, "metadata": {}, "source": [ "11.1 Class Predictions" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Classification models usually generate two types of predictions. Like regression models, classification models produce a continuous valued prediction, which is usually in the form of a probability (i.e., the predicted values of class membership for any individual sample are between 0 and 1 and sum to 1). In addition to a continuous prediction, classification models generate a predicted class, which comes in the form of a discrete category. For most practical applications, a discrete category prediction is required in order to make a decision.\n", "\n", "Although classification models produce both of these types of predictions, often the focus is on the discrete prediction rather than the continuous prediciton. However, the probability estimates for each class can be very useful for gauging the model's confidence about the predicted classification. For example, an email message with a predicted probability of being spam of 0.51 would be classified the same as a message with a predicted probability of being spam of 0.99. While both messages would be treated the same by the filter, we would have more confidence that the second message was truly spam. \n", "\n", "In some applications, the desired outcome is the predicted class probabilities which are then used as inputs for other calculations. Consider an insurance company that wants to uncover and prosecute fraudulent claims. Using historical claims data, a classification model could be built to predict the probability of claim fraud. This probability would then be combined with the company's investigation costs and potential monetary loss to determine if pursuing the investigation is in the best financial interest of the insurance company.\n", "\n", "As mentioned, most classification models generate predicted class probabilities. However, when some models are used for classification, like neural networks and partial least squares, they produce continuous predictions that do not follow the definition of a probability. For example, a partial least squares classification model would ceate 0/1 dummy variables for each class and simultaneously model these values as a function of the predictors. When samples are predicted, the model predictions are not guaranteed to be within 0 and 1. For classification models like these, a transformation must be used to coerce the predictions into \"probability-like\" values so that they can be interpreted and used for classification. One such method is the *softmax transformation* \n", "$$\\hat{p}_l^{*} = {e^{\\hat{y}_l} \\over \\sum_{j=1}^C e^{\\hat{y}_j}}$$\n", "where $\\hat{y}_l$ is the numeric model prediciton for the $l^{\\text{th}}$ class and $\\hat{p}_l^{*}$ is the transformed value between 0 and 1. To be clear, no probability statement is being created by this transformation; it merely ensures that the predictions have the same mathematical qualities as probabilities." ] }, { "cell_type": "heading", "level": 3, "metadata": {}, "source": [ "Well-Calibrated Probabilities" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Whether a classification model is used to predict spam email, or as inputs to insurance fraud, we desire that the estimated class probabilities are reflective of the true underlying probability of the sample. That is, the predicted class probability needs to be well-calibrated. To be well-calibrated, the probabilities must effectively reflect the true likelihood of the event of interest.\n", "\n", "One way to assess the quality of the class probabilities is using a calibration plot. For a given set of data, this plot shows some measure of the observed probability of an event versus the predicted class probability. One approach for creating this visualization is to score a collection of samples with known outcomes (preferably a test set) using a classification model. The next step is to bin the data into groups based on their class probabilities. For example, a set of bins might be [0, 10%], (10%, 20%], ..., (90%, 100%]. For each bin, determine the observed event rate. Suppose that 50 samples fell into the bin for class probabilities less than 10% and there was a single event. The midpoint of the bin is 5% and the observed event rate would be 2%. The calibration plot would display the midpoint of the bin on the x-axis and the observed event rate on the y-axis. If the points fall along a $45^{\\circ}$ line, the model has produced well-calibrated probabilities." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "To illustrate, a data set was simulated. For classes (1 and 2) and two predictors (A and B), the true probability (p) of the event is generated from the equation $$log({p \\over 1 - p}) = -1 - 2A - 0.2A^2 + 2B^2$$" ] }, { "cell_type": "code", "collapsed": false, "input": [ "# simulate two data sets\n", "def simu_prob(A, B):\n", " return np.exp(-1 - 2*A - 0.2*A**2 + 2*B**2)/(1 + np.exp(-1 - 2*A - 0.2*A**2 + 2*B**2))\n", "\n", "np.random.seed(3)\n", "\n", "X_train = np.random.multivariate_normal(mean = [1.0, 0.0], cov = [[1.0, 0.7], [0.7, 2.0]], size = 500)\n", "Y_train = np.array([np.random.binomial(1, simu_prob(x[0], x[1])) for x in X_train])" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 2 }, { "cell_type": "code", "collapsed": false, "input": [ "colors = ['b', 'r']\n", "markers = ['s', 'o']\n", "\n", "for idx, i in enumerate(Y_train):\n", " plt.scatter(X_train[idx, 0], X_train[idx, 1], \n", " c = colors[i], marker= markers[i], \n", " alpha = 0.4, s = 36)\n", "plt.xlabel(\"Predictor A\")\n", "plt.ylabel(\"Predictor B\")\n", "\n", "# solid black line denotes the 50% probability contour\n", "delta = 0.025\n", "x = np.arange(-3.0, 5.0, delta)\n", "y = np.arange(-3.0, 5.0, delta)\n", "X, Y = np.meshgrid(x, y)\n", "Z = simu_prob(X, Y)\n", "plt.contour(x, y, Z, levels = [0.5])" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "pyout", "prompt_number": 3, "text": [ "" ] }, { "metadata": {}, "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAlwAAAHNCAYAAADLzqXLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", "text": [ "" ] } ], "prompt_number": 3 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Two models were fit to the training set: quadratic discriminant analysis and a random forest model. A test set of $n = 1000$ samples was used to score the model and create the calibration plot." ] }, { "cell_type": "code", "collapsed": false, "input": [ "np.random.seed(5)\n", "\n", "X_test = np.random.multivariate_normal(mean = [1.0, 0.0], cov = [[1.0, 0.7], [0.7, 2.0]], size = 1000)\n", "Y_test = np.array([np.random.binomial(1, simu_prob(x[0], x[1])) for x in X_test])" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 4 }, { "cell_type": "code", "collapsed": false, "input": [ "from sklearn.qda import QDA\n", "from sklearn.ensemble import RandomForestClassifier\n", "\n", "qda = QDA()\n", "qda.fit(X_train, Y_train)\n", "\n", "print \"Predictive accuracy of QDA: {0}\".format(qda.score(X_test, Y_test))\n", "\n", "rf = RandomForestClassifier(n_estimators=1000, max_depth=5)\n", "rf.fit(X_train, Y_train)\n", "print \"Predictive accuracy of Random Forest: {0}\".format(rf.score(X_test, Y_test))" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "stream", "stream": "stdout", "text": [ "Predictive accuracy of QDA: 0.865\n", "Predictive accuracy of Random Forest: 0.864" ] }, { "output_type": "stream", "stream": "stdout", "text": [ "\n" ] } ], "prompt_number": 5 }, { "cell_type": "code", "collapsed": false, "input": [ "qda_predict = qda.predict_proba(X_test)[:, 1]\n", "rf_predict = rf.predict_proba(X_test)[:, 1]\n", "\n", "# calibration plot\n", "# predict_prob = simu_prob(X_test[:, 0], X_test[:, 1])\n", "calib_counts = np.zeros([2, 10])\n", "\n", "for i in range(0, 10):\n", " calib_counts[0, i] = np.mean(Y_test[(qda_predict > i*0.1) * (qda_predict < (i+1)*0.1)])\n", " calib_counts[1, i] = np.mean(Y_test[(rf_predict > i*0.1) * (rf_predict < (i+1)*0.1)])\n", " \n", "plt.plot(np.arange(0.05, 1.0, 0.1), calib_counts[0, :], 'x-', label = 'qda')\n", "plt.plot(np.arange(0.05, 1.0, 0.1), calib_counts[1, :], 'o-', label = 'rf')\n", "plt.xlabel('Bin Midpoint')\n", "plt.ylabel('Observed Event Percentage')\n", "plt.legend(loc = 'upper center', ncol = 2)\n", "plt.plot([0, 1], [0, 1], 'k--', alpha = 0.5)" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "pyout", "prompt_number": 6, "text": [ "[]" ] }, { "metadata": {}, "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAmUAAAHSCAYAAACtqLx3AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", "text": [ "" ] } ], "prompt_number": 6 }, { "cell_type": "markdown", "metadata": {}, "source": [ "The calibration plot shows that the QDA class probabilities tend to perform poorly compared to the random forest model. For example, in the bin with class probabilities ranging from 20% to 30%, the observed percentage of events for QDA was far lower than the percentage in the random forest model." ] }, { "cell_type": "heading", "level": 3, "metadata": {}, "source": [ "Presenting Class Probabilities" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Visualizations of the class probabilities are an effective method of communicating model results. For two classes, histograms of the predicted classes for each of the true outcomes illustrate the strengths and weaknesses of a model. In Chapter 4, we introduced the credit scoring example. Two classification models were created to predict the quality of a customer's credit: a support vector machine (SVM) and logistic regression. Since the performance of the two models were roughly equivalent, the logistic regression model was favored due to its simplicity." ] }, { "cell_type": "code", "collapsed": false, "input": [ "germancredit = pd.read_csv(\"../datasets/GermanCredit/GermanCredit.csv\")\n", "germancredit.drop('Unnamed: 0', axis=1, inplace = True) # drop the first column\n", "germancredit.head(5)" ], "language": "python", "metadata": {}, "outputs": [ { "html": [ "
\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
DurationAmountInstallmentRatePercentageResidenceDurationAgeNumberExistingCreditsNumberPeopleMaintenanceTelephoneForeignWorkerClass...OtherInstallmentPlans.BankOtherInstallmentPlans.StoresOtherInstallmentPlans.NoneHousing.RentHousing.OwnHousing.ForFreeJob.UnemployedUnskilledJob.UnskilledResidentJob.SkilledEmployeeJob.Management.SelfEmp.HighlyQualified
0 6 1169 4 4 67 2 1 0 1 Good... 0 0 1 0 1 0 0 0 1 0
1 48 5951 2 2 22 1 1 1 1 Bad... 0 0 1 0 1 0 0 0 1 0
2 12 2096 2 3 49 1 2 1 1 Good... 0 0 1 0 1 0 0 1 0 0
3 42 7882 2 4 45 1 2 1 1 Good... 0 0 1 0 0 1 0 0 1 0
4 24 4870 3 4 53 2 2 1 1 Bad... 0 0 1 0 0 1 0 0 1 0
\n", "

5 rows \u00d7 62 columns

\n", "
" ], "metadata": {}, "output_type": "pyout", "prompt_number": 7, "text": [ " Duration Amount InstallmentRatePercentage ResidenceDuration Age \\\n", "0 6 1169 4 4 67 \n", "1 48 5951 2 2 22 \n", "2 12 2096 2 3 49 \n", "3 42 7882 2 4 45 \n", "4 24 4870 3 4 53 \n", "\n", " NumberExistingCredits NumberPeopleMaintenance Telephone ForeignWorker \\\n", "0 2 1 0 1 \n", "1 1 1 1 1 \n", "2 1 2 1 1 \n", "3 1 2 1 1 \n", "4 2 2 1 1 \n", "\n", " Class ... OtherInstallmentPlans.Bank OtherInstallmentPlans.Stores \\\n", "0 Good ... 0 0 \n", "1 Bad ... 0 0 \n", "2 Good ... 0 0 \n", "3 Good ... 0 0 \n", "4 Bad ... 0 0 \n", "\n", " OtherInstallmentPlans.None Housing.Rent Housing.Own Housing.ForFree \\\n", "0 1 0 1 0 \n", "1 1 0 1 0 \n", "2 1 0 1 0 \n", "3 1 0 0 1 \n", "4 1 0 0 1 \n", "\n", " Job.UnemployedUnskilled Job.UnskilledResident Job.SkilledEmployee \\\n", "0 0 0 1 \n", "1 0 0 1 \n", "2 0 1 0 \n", "3 0 0 1 \n", "4 0 0 1 \n", "\n", " Job.Management.SelfEmp.HighlyQualified \n", "0 0 \n", "1 0 \n", "2 0 \n", "3 0 \n", "4 0 \n", "\n", "[5 rows x 62 columns]" ] } ], "prompt_number": 7 }, { "cell_type": "code", "collapsed": false, "input": [ "germancredit.groupby('Class').count()" ], "language": "python", "metadata": {}, "outputs": [ { "html": [ "
\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
DurationAmountInstallmentRatePercentageResidenceDurationAgeNumberExistingCreditsNumberPeopleMaintenanceTelephoneForeignWorkerCheckingAccountStatus.lt.0...OtherInstallmentPlans.BankOtherInstallmentPlans.StoresOtherInstallmentPlans.NoneHousing.RentHousing.OwnHousing.ForFreeJob.UnemployedUnskilledJob.UnskilledResidentJob.SkilledEmployeeJob.Management.SelfEmp.HighlyQualified
Class
Bad 300 300 300 300 300 300 300 300 300 300... 300 300 300 300 300 300 300 300 300 300
Good 700 700 700 700 700 700 700 700 700 700... 700 700 700 700 700 700 700 700 700 700
\n", "

2 rows \u00d7 61 columns

\n", "
" ], "metadata": {}, "output_type": "pyout", "prompt_number": 8, "text": [ " Duration Amount InstallmentRatePercentage ResidenceDuration Age \\\n", "Class \n", "Bad 300 300 300 300 300 \n", "Good 700 700 700 700 700 \n", "\n", " NumberExistingCredits NumberPeopleMaintenance Telephone \\\n", "Class \n", "Bad 300 300 300 \n", "Good 700 700 700 \n", "\n", " ForeignWorker CheckingAccountStatus.lt.0 ... \\\n", "Class ... \n", "Bad 300 300 ... \n", "Good 700 700 ... \n", "\n", " OtherInstallmentPlans.Bank OtherInstallmentPlans.Stores \\\n", "Class \n", "Bad 300 300 \n", "Good 700 700 \n", "\n", " OtherInstallmentPlans.None Housing.Rent Housing.Own Housing.ForFree \\\n", "Class \n", "Bad 300 300 300 300 \n", "Good 700 700 700 700 \n", "\n", " Job.UnemployedUnskilled Job.UnskilledResident Job.SkilledEmployee \\\n", "Class \n", "Bad 300 300 300 \n", "Good 700 700 700 \n", "\n", " Job.Management.SelfEmp.HighlyQualified \n", "Class \n", "Bad 300 \n", "Good 700 \n", "\n", "[2 rows x 61 columns]" ] } ], "prompt_number": 8 }, { "cell_type": "code", "collapsed": false, "input": [ "# use stratified sampling to split training and test sets\n", "def stratified_sampling(data, target, prop_train):\n", " '''a stratified random sampling based on proportion of the target outcomes.'''\n", " from collections import Counter\n", " \n", " n_class = dict(Counter(data[target]).items()) # frequency of possible outcomes\n", " data.reindex(np.random.permutation(data.index)) # random shuffle\n", " for key, val in n_class.iteritems():\n", " n_train = np.int(val*prop_train)\n", " try:\n", " trainset = trainset.append(data[data[target] == key].iloc[:n_train,:], ignore_index=True)\n", " testset = testset.append(data[data[target] == key].iloc[n_train:, :], ignore_index=True)\n", " except NameError:\n", " trainset = data[data[target] == key].iloc[:n_train,:]\n", " testset = data[data[target] == key].iloc[n_train:, :]\n", " \n", " return trainset, testset\n", " \n", "credit_train, credit_test = stratified_sampling(germancredit, 'Class', 0.8)\n", "print \"The size of training set is {0} and test set is {1}.\".format(credit_train.shape[0], credit_test.shape[0])" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "stream", "stream": "stdout", "text": [ "The size of training set is 800 and test set is 200.\n" ] } ], "prompt_number": 9 }, { "cell_type": "code", "collapsed": false, "input": [ "X_train = credit_train.drop('Class', axis=1)\n", "y_train = credit_train['Class']\n", "X_test = credit_test.drop('Class', axis=1)\n", "y_test = credit_test['Class']" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 10 }, { "cell_type": "code", "collapsed": false, "input": [ "# fit a logistic regression\n", "from sklearn.linear_model import LogisticRegression\n", "\n", "lreg = LogisticRegression()\n", "lreg.fit(X_train, y_train)\n", "\n", "lreg_pred = lreg.predict(X_test)\n", "lreg_pred_prob = lreg.predict_proba(X_test)" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 11 }, { "cell_type": "code", "collapsed": false, "input": [ "fig, (ax1, ax2) = plt.subplots(ncols = 2, sharey=True)\n", "\n", "ax1.hist(lreg_pred_prob[np.where(y_test == 'Bad'), 0].T, 20)\n", "ax1.set_title('True Outcome: Bad Credit')\n", "\n", "ax2.hist(lreg_pred_prob[np.where(y_test == 'Good'), 0].T, 20)\n", "ax2.set_title('True Outcome: Good Credit')\n", "\n", "fig.text(0.5, 0.06, 'Probability of Bad Credit', ha='center', va='center')" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "pyout", "prompt_number": 12, "text": [ "" ] }, { "metadata": {}, "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAlIAAAHnCAYAAABzMLMoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", "text": [ "" ] } ], "prompt_number": 12 }, { "cell_type": "markdown", "metadata": {}, "source": [ "The probability of bad credit for the customers with good credit shows a skewed distribution where most customers' probabilities are quite low. In contrast, the probabilities for the customers with bad credit are flat (or uniformly distributed), reflecting the model's inability to distinguish bad credit cases." ] }, { "cell_type": "code", "collapsed": false, "input": [ "predict_counts = np.zeros([10])\n", "\n", "for i in range(10):\n", " predict_counts[i] = np.mean(y_test[(lreg_pred_prob[:, 0] > i*0.1) * (lreg_pred_prob[:, 0] < (i+1)*0.1)] == 'Bad')\n", " \n", "plt.plot(np.arange(0.05, 1.0, 0.1), predict_counts)\n", "plt.plot([0, 1], [0, 1], 'k--', alpha = 0.5)\n", "plt.xlabel(\"Bin Midpoint\")\n", "plt.ylabel(\"Observed Event Percentage\")" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "pyout", "prompt_number": 13, "text": [ "" ] }, { "metadata": {}, "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAmUAAAHSCAYAAACtqLx3AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", "text": [ "" ] } ], "prompt_number": 13 }, { "cell_type": "markdown", "metadata": {}, "source": [ "The calibration plot shows the accuracy of the probability of bad credit degrades as it becomes larger to the point where a few samples with bad credit were predicted with a probability above 80%. This pattern is indicative of a model that has both poor calibration and poor performance.\n", "\n", "When there are three or more classes, a heat map of the class probabilities can help gauge the confidence in the predictions." ] }, { "cell_type": "heading", "level": 3, "metadata": {}, "source": [ "Equivocal Zones" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "An approach to improving classification performance is to create an equivocal or indeterminate zone where the class is not formally predicted when the confidence is not high. For a two-class problem that is nearly balanced in the response, the equivocal zone could be defined as $0.50 \\pm z$. If z were 0.10, then samples with prediction probabilities between 0.40 and 0.60 would be called \"equivocal\". In this case, model performance would be calculated excluding the samples in the indeterminate zone. For data sets with more than 2 classes (C > 2), similar thresholds can be applied where the largest class probability must be larger than (1/C + z) to make a definitive prediction." ] }, { "cell_type": "heading", "level": 2, "metadata": {}, "source": [ "11.2 Evaluating Predicted Classes" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "A common method for describing the performance of a classification model is the *confusion matrix*. This is a simple cross-tabulation of the observed and predicted classes for the data.\n", "\n", "Predict\\Observe | Event | Non-event\n", "------------------|-------|----------\n", "Event | TP | FP\n", "Non-event | FN | TN\n", "\n", "The above table shows an example when the outcome has two classes. Diagonal cells denote cases where the classes are correctly predicted while the off-diagonals illustrate the number of errors for each possible case.\n", "\n", "The simplest metric is the overall accuracy rate. This reflects the agreement between the observed and predicted classes and has the most straightforward interpretation. However, there are a few disadvantages to using this statistic. First, overall accuracy counts make no distinction about the type of errors being made. In situations where the costs are different, accuracy may not measure the important model characteristics.\n", "\n", "Second, one must consider the natural frequencies of each class. For example, in the USA, pregnant women routinely have blood drawn for alpha-fetoprotein testing, which attempts to detect genetic problems such as Down syndrome. Suppose the rate of this disorder in fetuses is approximately 1 in 800 or about one-tenth of on percent. A predictive model can achieve almost perfect accuracy by predicting all samples to be negative for Down syndrome.\n", "\n", "What benchmark accuracy rate should be used to determine whether a model is performing adequantely? The no-information rate is the accuracy rate that can be achieved without a model. There are various ways to define this rate. For a data set with $C$ classes, the simplest definition, based on pure randomness, is $1/C$. However, this does not take into account the relative frequencies of the classes in the training set. An alternate definition of the no-information rate is the percentage of the largest class in the training set. Models with accuracy greater than this rate might be considered reasonable. The effect of severe class imbalances and some possible remedies are discussed later.\n", "\n", "Rather than calculate the overall accuracy and compare it to the no-information rate, other metrics can be used that take into account the class distribution of the training set samples. The *Kappa statistic* takes into account the accuracy that would be generate simply by chance. $$Kappa = {O - E \\over 1 - E}$$\n", "where $O$ is the observed accuracy and $E$ is the expected accuracy based on the marginal totals of the confusion matrix. The statistic can take on values between -1 and 1; a value of 0 means there is no aggreement between the observed and predicted classes. Negative values indicate that the prediction is in the *opposite* direction of the truth, but large negative values seldom occur, if ever, when working with predictive models. When the class distributions are equivalent, overall accuracy and Kappa are proportional. Depending on the context, Kappa values within 0.3 to 0.5 indicate reasonable aggreement. Suppose the accuracy for a model is high (90%) but the expected accuracy is also high (85%), the Kappa statistic would show moderate aggreement (Kappa = 1/3) between the observed and predicted classes." ] }, { "cell_type": "heading", "level": 3, "metadata": {}, "source": [ "Two-Class Problems" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Consider the cases where there are two classes.\n", "\n", "Predict\\Observe | Event | Non-event\n", "------------------|-------|----------\n", "Event | TP | FP\n", "Non-event | FN | TN\n", "\n", "The table shows the confusion matrix for generic classes \"event\" and \"nonevent\". The top row of the table corresponds to samples predicted to be events. Some are predicted correctly (the true positives, TP) while others are inaccurately classified (the false positives, FP). Similarly, the second row contains the predicted negatives with the true negatives (TN) and false nagatives (FN).\n", "\n", "For two classes, there are additional statistics that may be relevant when one class is interpreted as the event of interest. The *sensitivity* of the model is the rate that the event of interest is predicted correctly for all samples having the event, i.e., \n", "$$Sensitivity = {\\# \\text{samples with the event and predicted to have the event} \\over \\# \\text{samples having the event}} = {TP \\over TP + FN}$$\n", "The sensitivity is sometimes considered the *true positive rate* since it measures the accuracy in the event population. Conversely, the *specificity* is defined as the rate that nonevent samples are predicted as nonevents, i.e.\n", "$$Specificity = {\\# \\text{samples without the event and predicted as non-events} \\over \\# \\text{samples without the event}} = {TN \\over TN + FP}$$\n", "The *false positive rate* is defined as one minus the specificity. Assuming a fixed level of accuracy for the model, there is typically a trade-off to be made between the sensitivity and specificity. Intuitively, increasing the sensitivity of a model is likely to incur a loss of specificity, since more samples are being predicted as events. Potential trade-offs between sensitivity and specificity may be appropriate when there are different penalties associated with each type of error. The *receiver operating characteristic (ROC) curve* is one technique for evaluating this trade-off." ] }, { "cell_type": "code", "collapsed": false, "input": [ "# pay a revisit to the credit data set\n", "from sklearn.metrics import confusion_matrix\n", "confusion_matrix(y_test, lreg_pred)" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "pyout", "prompt_number": 14, "text": [ "array([[ 31, 29],\n", " [ 25, 115]])" ] } ], "prompt_number": 14 }, { "cell_type": "code", "collapsed": false, "input": [ "from __future__ import division\n", "\n", "print \"The overall accuracy: {0} and the no-information rate: {1}\".format((31+115)/200, np.mean(y_test == 'Good'))\n", "rnd_aggreement = ((31+29)/(200)) * ((31+25)/200) + ((25+115)/200)*((29+115)/200)\n", "print \"Kappa = {0:.4f}\".format(((31+115)/200 - rnd_aggreement)/(1 - rnd_aggreement))\n", "print \"Sensitivity = {0:.4f}\".format(31/(31+25))\n", "print \"Specificity = {0:.4f}\".format(115/(115+29))" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "stream", "stream": "stdout", "text": [ "The overall accuracy: 0.73 and the no-information rate: 0.7\n", "Kappa = 0.3447\n", "Sensitivity = 0.5536\n", "Specificity = 0.7986\n" ] } ], "prompt_number": 15 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Clearly, the model has trouble predicting when customers have bad credit. This is likely due to the imbalance of the classes and a lack of a strong predictor for bad credit." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Often, there is interest in having a single measure that reflects the false-positive and false-negative rates. Youden's J Index $$J = Sensitivity + Specificity - 1$$ measures the proportions of correctly predicted samples for both the event and non-event groups. In some contexts, this may be an appropriate method for summarizing the magnitude of both types of errors. The most common method for combining sensitivity and specificity into a single value uses the receiver operating characteristic (ROC) curve.\n", "\n", "One often overlooked aspect of sensitivity and specificity is that they are conditional measures. Sensitivity is the accuracy rate for only the event population (and specificity for the non-events). Using the sensitivity and specificity, the obstetrician can make statements such as \"assuming that the fetus does not have Down syndrome, the test has an accuracy of 95%.\" However, these statements might not be helpful to a patient since, for a new samples, all that known is the prediction. The person using the model prediction is typically interested in unconditional queries such as \"what are the chances that the fetus has the genetic disorder?\" This depends on three values: the sensitivity and specificity of the diagnostic test and the prevalence of the event in the population. Intuitively, if the event is rare, this should be reflected in the answer. Taking the prevalence into account, the analog to specificity is the *positive predicted value*, and the analog to specificity is the *negative predicted value*. These values make unconditional evaluations of the data. The positive predicted value answers the question \"what is the probability that this sample is an event?\" The formulas are \n", "$$PPV = {Sensitivity \\times Prevalence \\over (Sensitivity \\times Prevalence) + ((1 - Specificity) \\times (1 - Prevalence))}$$\n", "$$NPV = {Specificity \\times (1 - Prevalence) \\over (Prevalence \\times (1 - Sensitivity)) + (Specificity \\times (1 - Prevalence))}$$\n", "\n", "Clearly, the predictive values are nontrival combinations of performance and the rate of events. Predictive values are not often used to characterize the model. There are several reasons why, most of which are related to prevalence. First, prevalence is hard to quantify. Also, the prevalence is dynamic." ] }, { "cell_type": "heading", "level": 3, "metadata": {}, "source": [ "Non-Accuracy-Based Criteria" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "For many commercial applications of predictive models, accuracy is not the primary goal for the model. Often, the purpose of the model might be to:\n", "- Predict investment opportunities that maximize return\n", "- Improve customer satisfaction by market segmentation\n", "- Lower inventory costs by improving product demand forecasts\n", "- Reduce costs associated with fraudulent transactions\n", "\n", "While accuracy is important, it only describes how well the model predicts the data. If the model is fit for purpose, other more direct metrics of performance should be considered. These metrics quantify the consequences of correct and incorrect predictions." ] }, { "cell_type": "heading", "level": 2, "metadata": {}, "source": [ "11.3 Evaluating Class Probabilities" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Class probabilities potentially offer more information about model predictions than the simple class value." ] }, { "cell_type": "heading", "level": 3, "metadata": {}, "source": [ "Receiver Operating Characteristic (ROC) Curves" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "ROC curves were designed as a general method that, given a collection of continuous data points, determine an effective threshold such that values above the threshold are indicative of a specific event.\n", "\n", "The ROC curve is created by evaluating the class probabilities for the model across a continuum of thresholds. For each candidate threshold, the resulting true-positive rate (i.e. the sensitivity) and the false-positive rate (one minus the specificity) are plotted against each other." ] }, { "cell_type": "code", "collapsed": false, "input": [ "# ROC curve for credit data\n", "from sklearn.metrics import roc_curve, auc\n", "\n", "fpr, tpr, thres = roc_curve(y_test, lreg_pred_prob[:, 0], pos_label= 'Bad')\n", "roc_auc = auc(fpr, tpr)\n", "\n", "plt.figure()\n", "plt.plot(fpr, tpr, label='ROC curve (area = %0.2f)' % roc_auc)\n", "\n", "# plot specific value based on threshold\n", "fpr_50 = fpr[np.where(thres < 0.5)[0][0]]\n", "tpr_50 = tpr[np.where(thres < 0.5)[0][0]]\n", "plt.plot(fpr_50, tpr_50, 'o')\n", "plt.text(fpr_50+0.02, tpr_50-0.01, \n", " \"0.500 (Spec = {0:.3f}, Sens = {1:.3f})\".format(1-fpr_50, tpr_50))\n", "\n", "fpr_30 = fpr[np.where(thres < 0.3)[0][0]]\n", "tpr_30 = tpr[np.where(thres < 0.3)[0][0]]\n", "plt.plot(fpr_30, tpr_30, 'o')\n", "plt.text(fpr_30+0.02, tpr_30-0.01, \n", " \"0.300 (Spec = {0:.3f}, Sens = {1:.3f})\".format(1-fpr_30, tpr_30))\n", "\n", "plt.plot([0, 1], [0, 1], 'k--')\n", "plt.xlim([0.0, 1.0])\n", "plt.ylim([0.0, 1.05])\n", "plt.xlabel('1 - Specificity (False Positive Rate)')\n", "plt.ylabel('Sensitivity (True Positive Rate)')\n", "plt.title('Receiver operating characteristic example')\n", "plt.legend(loc=\"lower right\")\n", "plt.show()" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAmUAAAHcCAYAAACXot0HAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n", "text": [ "" ] } ], "prompt_number": 16 }, { "cell_type": "markdown", "metadata": {}, "source": [ "The green point is the default 50% threshold while the red point corresponds to the performance characteristics for a threshold of 30%. In this figure, the numbers in parentheses are $(specificity, sensitivity)$. Note that the trajectory of the curve between (0, 0) and the 50% threshold is steep, indicating that the sensitivity is increasing at a greater rate than the decrease in specificity. However, when the sensitivity is greater than 70%, there is a more significant decrease in specificity than the gain in sensitivity.\n", "\n", "This plot is a helpful tool for choosing a threshold that appropriately maximizes the trade-off between sensitivity and specificity. However, altering the threshold only has the effect of making the samples more positive (or negative). In the confusion matrix, it cannot move samples out of both off-diagnoal table cells. There is almost always a decrease in either sensitivity or specificity.\n", "\n", "The ROC curve can also be used for a quantitative assessment of the model. A perfect model that completely separates the two classes would have 100% sensitivity and specificity. Graphically, the ROC curve would be a single step between (0, 0) nad (0, 1) and remain constant from (0, 1) and (1, 1). The area under the ROC curve for such a model would be one. A completely ineffective model would result in an ROC curve that closely follows the $45^{\\circ}$ diagnoal line and would have an area under the ROC curve of approximately 0.5. To visually compare different models, their ROC curves can be superimposed on the same graph. Comparing ROC curves can be useful in contrasting two or more models with different predictor sets (for the same model), differnt tuning parameters (within model comparisons), or complete different classifiers (between models).\n", "\n", "The optimal model should be shifted towards the upper left corner of the plot. Alternatively, the model with the largest area under the ROC curve would be the most effective. \n", "\n", "One advantage of using ROC curves to characterize models is that, since it is a function of sensitivity and specificity, the curve is insensitive to disparities in the class proportions. A disadvantage of using the area under the curve to evaluate models is that it obscures information. When comparing models, it is common that no individual ROC curve is uniformly better than another. By summarizing these curves, there is a loss of information, especially if one particular area of the curve is of interest. The partial area under the ROC curve is an alternative that focuses on specific parts of the curve." ] }, { "cell_type": "heading", "level": 3, "metadata": {}, "source": [ "Life Charts" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Lift charts are a visualization tool for assessing the ability of a model to detect events in a data set with two classes. Suppose a group of samples with M events is scored using the event class probability. When ordered by the class probability, one would hope that the events are ranked higher than the non-events. Lift charts do just this: rank the samples by their scores and determine the cumulative event rate as more samples are evaluated. In the optimal case, the M highest-ranked samples would contain all M events. When the model is non-informative, the highest ranked X% of the data would contain, on average, X events. The *lift* is the number of samples detected by a model above a completely random selection of samples.\n", "\n", "To construct the *lift charts* we would take the following steps:\n", "1. Predict a set of samples that were not used in the model building process but have known outcomes.\n", "2. Determine the baseline event rate, i.e. the percent of true events in the entire data set.\n", "3. Order the data by the classification probability of the event of interest.\n", "4. For each unique class probability value, calculate the percent of true events in all samples below the probability value.\n", "5. Divide the percent of true events for each probability threshold by the baseline event rate.\n", "\n", "The lift chart plots the cumulative gain/lift against the cumulative percentage of samples that have been screened. With a 50% event arte, the non-informative model has a curve that is close to the $45^{\\circ}$ reference line, meaning that the model has no benefit for ranking samples." ] }, { "cell_type": "code", "collapsed": false, "input": [ "# lift chart for credit data\n", "y_test_label = np.array(map(lambda c: 1 if c == 'Bad' else 0, y_test)).reshape(len(y_test), 1)\n", "lift_curve = np.hstack((y_test_label, lreg_pred_prob[:, 0].reshape(len(y_test), 1)))\n", "lift_curve = lift_curve[lift_curve[:, 1].argsort()[::-1]]\n", "\n", "plt.plot(np.arange(1.0, len(y_test)+1)/len(y_test), \n", " np.cumsum(lift_curve[:, 0])/np.sum(lift_curve[:, 0]),\n", " label = 'Logistic Regression')\n", "plt.fill_between(np.arange(1.0, len(y_test)+1)/len(y_test), \n", " 1.0*np.cumsum(y_test_label)/np.sum(y_test_label),\n", " 1.0*np.cumsum(np.random.permutation(y_test_label))/np.sum(y_test_label),\n", " facecolor = 'gray', alpha = 0.5)\n", "plt.legend(loc = 4)\n", "plt.xlabel('% Samples Tested')\n", "plt.ylabel('% Events Found')" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "pyout", "prompt_number": 17, "text": [ "" ] }, { "metadata": {}, "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAmUAAAHSCAYAAACtqLx3AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nqwZJPEZnAAAAAElFTkSuQmCC\n", "text": [ "" ] } ], "prompt_number": 17 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Note that, since the event rate of the credit data is not balanced, the non-informative boundary is above the $45^{\\circ}$ line.\n", "\n", "Like ROC curves, the lift curves for different models can be compared to find the most appropriate model and the area under the curve can be used as a quantitative measure of performance. Also like ROC curves, some parts of the lift curve are of more interest than others. For example, the section of the curve associated with the highest-ranked samples should have an enriched true-positive rate and is likely to be the most important part of the curve." ] } ], "metadata": {} } ] }