2*classifier correlation-based entropy-based mRMR-MIQ mRMR-MID F-score on   size F-score size F-score size F-score size F-score all features SVM-linear 40 0.750 37 0.750 22 0.757 20 0.752 0.726 SVM-poly 45 0.762 47 0.759 22 0.761 23 0.758 0.737 LR 31 0.768 27 0.766 23 0.774 20 0.763 0.769 ANN 31 0.732 72 0.769 23 0.727 86 0.770 0.777 LDA 37 0.772 30 0.769 22 0.770 31 0.766 0.755 KNN 29 0.729 28 0.723 29 0.747 29 0.743 0.639 C45 55 0.707 56 0.703 14 0.715 49 0.702 0.714 NB 36 0.748 45 0.749 33 0.751 31 0.753 0.739

Table 5: Optimal feature set and performance measured by F-score. Elapsed time is measured based on 10-fold cross validation.