In addition to using the boosting function in the adabag package, one can also use the caret package to perform a classification with the boosting method:

1. First, load the mboost and pROC package:

        > library(mboost)
        > install.packages("pROC")
        > library(pROC)  

2. We can then set the training control with the trainControl function and use the train function to train the classification model with AdaBoost.M1:

        > set.seed(2)
        > ctrl = trainControl(method = "repeatedcv", repeats = 1, class
        Probs = TRUE, summaryFunction = twoClassSummary)
        > ada.train = train(churn ~ ., data = trainset, method = "ada", 
        metric = "ROC", trControl = ctrl)  

3. Use the summary function to obtain the details of the classification model:

        > ada.train$result
        Output
        nu maxdepth iter       ROC      Sens        Spec      ROCSD
        SensSD      SpecSD
        1 0.1        1   50 0.8571988 0.9152941 0.012662155 0.03448418
        0.04430519 0.007251045
        4 0.1        2   50 0.8905514 0.7138655 0.006083679 0.03538445
        0.10089887 0.006236741
        7 0.1        3   50 0.9056456 0.4036134 0.007093780 0.03934631
        0.09406015 0.006407402
        2 0.1        1  100 0.8550789 0.8918487 0.015705276 0.03434382
        0.06190546 0.006503191
        5 0.1        2  100 0.8907720 0.6609244 0.009626724 0.03788941
        0.11403364 0.006940001
        8 0.1        3  100 0.9077750 0.3832773 0.005576065 0.03601187
        0.09630026 0.003738978
        3 0.1        1  150 0.8571743 0.8714286 0.016720505 0.03481526
        0.06198773 0.006767313
        6 0.1        2  150 0.8929524 0.6171429 0.011654617 0.03638272
        0.11383803 0.006777465
        9 0.1        3  150 0.9093921 0.3743697 0.007093780 0.03258220
        0.09504202 0.005446136

4. Use the plot function to plot the ROC curve within different iterations:

        > plot(ada.train)

5. Finally, we can make predictions using the predict function and view the classification table:

        > ada.predict = predict(ada.train, testset, "prob")
        > ada.predict.result = ifelse(ada.predict[1] > 0.5, "yes", "no")
    
        > table(testset$churn, ada.predict.result)
        Output
             ada.predict.result
               no yes
          yes  40 101
          no  872   5