Figure 5.

The ROC curves for three classifiers using random over-sampling and under-sampling methods. Figure (a1) KStar with over-sampling method; (a2) KStar with under-sampling method; (b1) C4.5 with over-sampling method; (b2) C4.5 with under-sampling method; (c1) MLP with over-sampling method; (c2) MLP with under-sampling method. The blue and yellow lines in ROC curves represent the different threshold values which are used to generate ROC.