Figure 3.

ROC Curves of five different SVM models. A ROC curve provides a graphical representation of the relationship between the true-positive and false-positive prediction rate of a SVM model. ROC curve is obtained by plotting all 1-Specificity values (false-positive rate) on the X axis and Sensitivity (true-positive rate) on the Y axis. The resulting area under the ROC curve is an important index for evaluating the classification performance, i.e. the highest and leftmost ROC curve in the plot represents the best SVM model.