Table 4.

Performance attained by our classification scheme compared to that attained via random under-sampling and by the ensemble SVM classification method proposed by Fang et al. (2). Standard deviation is shown in parentheses. The highest performance level along each metric is shown in boldface

Method	Precision	Recall	F-measure	MCC
Random under-sampling	0.741 (0.004)	0.694 (0.006)	0.717 (0.004)	0.673 (0.005)
Ensemble SVM	0.692 (0.02)	0.642 (0.03)	0.662 (0.01)	0.613 (0.01)
Our final classifier (K-means, K = 5)	0.719 (0.008)	0.791 (0.012)	0.753 (0.004)	0.711 (0.004)