IMPORTING DATA

Reading Data

library(ISLR)
# reading inbuilt data as data frame
default.df <- as.data.frame(Default)
# attach data frame
attach(default.df)
# dimension of the data frame
dim(default.df)

[1] 10000     4

Data Structure

# structure of the data table
str(default.df)

'data.frame':   10000 obs. of  4 variables:
 $ default: Factor w/ 2 levels "No","Yes": 1 1 1 1 1 1 1 1 1 1 ...
 $ student: Factor w/ 2 levels "No","Yes": 1 2 1 1 1 2 1 2 1 1 ...
 $ balance: num  730 817 1074 529 786 ...
 $ income : num  44362 12106 31767 35704 38463 ...

Descriptive Statistics

# descriptive statistics of the dataframe
library(psych)
describe(default.df)[, c(1:5)]

         vars     n     mean       sd   median
default*    1 10000     1.03     0.18     1.00
student*    2 10000     1.29     0.46     1.00
balance     3 10000   835.37   483.71   823.64
income      4 10000 33516.98 13336.64 34552.64

SPLITTING DATA

(DATA TRAINING AND TESTING)

CONTROL PARAMETERS

Control Parameters

# control parameters
objControl <- trainControl(method = "boot", 
                           number = 2, 
                           returnResamp = 'none', 
                           summaryFunction = twoClassSummary, 
                           classProbs = TRUE,
                           savePredictions = TRUE)

Building Model using caret Package

# model building using caret package
set.seed(766)
caretLogitModel <- train(trainData.df[, 2:4],
                      trainData.df[, 1],
                      method = 'glmStepAIC',
                      trControl = objControl,
                      metric = "ROC",
                      verbose = FALSE)

Model Summary

# summary of the model
summary(caretLogitModel)

Call:
NULL

Deviance Residuals: 
    Min       1Q   Median       3Q      Max  
-2.4914  -0.1374  -0.0521  -0.0185   3.7862  

Coefficients:
              Estimate Std. Error z value Pr(>|z|)    
(Intercept) -1.100e+01  4.260e-01 -25.832  < 2e-16 ***
studentYes  -7.034e-01  1.655e-01  -4.249 2.15e-05 ***
balance      5.881e-03  2.662e-04  22.092  < 2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 2340.6  on 8000  degrees of freedom
Residual deviance: 1227.7  on 7998  degrees of freedom
AIC: 1233.7

Number of Fisher Scoring iterations: 8

Predicted Probabilities of Default (Yes/No), based on Test Data

# predicted probabilities
predTestProb <- predict(caretLogitModel, testData.df, type = "prob")
# plot of probabilities
plot(predTestProb[,2], 
     main = "Scatterplot of Probabilities of Default (test data)", 
     xlab = "Customer ID", ylab = "Predicted Probability of Default")

Prediction based on Test Data

# prediction of default = {no, yes} on test data
predClass <- predict(object = caretLogitModel, testData.df[, 2:4], type = 'raw')
table(predClass)

predClass
  No  Yes 
1976   23 

Confusion Matrix based on Test Data

# confusion matrix
confusionMatrix(predClass, testData.df$default, positive = "Yes")

Confusion Matrix and Statistics

          Reference
Prediction   No  Yes
       No  1925   51
       Yes    8   15

               Accuracy : 0.9705          
                 95% CI : (0.9621, 0.9775)
    No Information Rate : 0.967           
    P-Value [Acc > NIR] : 0.2098          

                  Kappa : 0.3256          
 Mcnemar's Test P-Value : 4.553e-08       

            Sensitivity : 0.227273        
            Specificity : 0.995861        
         Pos Pred Value : 0.652174        
         Neg Pred Value : 0.974190        
             Prevalence : 0.033017        
         Detection Rate : 0.007504        
   Detection Prevalence : 0.011506        
      Balanced Accuracy : 0.611567        

       'Positive' Class : Yes             

ROC Plot on the Test data

library(ROCR)
#Every classifier evaluation using ROCR starts with creating a prediction object. This function is used to transform the input data into a standardized format.
PredictObject <- prediction(predTestProb[2], testData.df$default)

# All kinds of predictor evaluations are performed using the performance function
PerformObject <- performance(PredictObject, "tpr","fpr")

# Plot the ROC Curve for Credit Card Default
plot(PerformObject, 
     main = "ROC Curve for CC Default",
     col = 2,
     lwd = 2)
abline(a = 0,b = 1,lwd = 2,lty = 3,col = "black")

Sensitivity - Specificity Plot based on the Test data

library(ROCR)

#Every classifier evaluation using ROCR starts with creating a prediction object. This function is used to transform the input data into a standardized format.
PredictObject <- prediction(predTestProb[2], testData.df$default)

# All kinds of predictor evaluations are performed using the performance function
PerformObject2 <- performance(PredictObject, "sens","spec")

# Plot the ROC Curve for Credit Card Default
plot(PerformObject2, 
     main = "Sensitivity - Specificity Plot for CC Default", 
     col = "blue", 
     lwd = 2)
abline(a = 1, b = -1, lwd = 2, lty = 3, col = "black")