Here, we demonstrate how we can compare fitted models by illustrating their ROC curve in one figure. First, we set up the control of the training process with a 10-fold cross validation in three repetitions with the performance evaluation in twoClassSummary. After setting up control of the training process, we then apply glm, svm, and rpart algorithms on the training dataset to fit the classification models. Next, we can make a prediction based on each generated model and plot the ROC curve, respectively. Within the generated figure, we find that the model trained by svm has the largest area under curve, which is 0.9233 (plotted in green), the AUC of the glm model (red) is 0.82, and the AUC of the rpart model (blue) is 0.7581. From Figure 7, it is obvious that svm performs the best among all the fitted models on this training dataset (without requiring tuning).