Increasing Minority Recall Support Vector Machine Model for Imbalanced Data Classification

Table 3

Recall and specificity (represented by r and s, respectively).

Dataset	No cost		Prorated cost		B-SVM		CS-SVM		BP		Proposed method
	r	s	r	s	r	s	r	s	r	s	r	s
Breast cancer o	0.983	0.921	0.749	0.891	0.975	0.958	0.988	0.958	0.988	0.969	1	0.930
Breast cancer d	0.390	0.939	0.872	0.861	0.920	0.936	0.920	0.927	0.925	0.964	1	0.701
Ecoli1	0.811	0.796	0.775	0.250	0.898	0.927	0.910	0.873	0.975	0.829	1	0.807
Ecoli2	0.982	0.230	0.742	0.732	0.865	0.932	0.942	0.937	0.793	0.835	1	0.823
Glass6	0.867	0.816	0.333	0.795	0.900	0.978	0.933	0.962	0.893	0.984	1	0.919
Car1v3	0.832	0.830	1.000	0.956	1.000	0.990	1.000	0.997	1.000	0.935	1	0.997
Car1v4	0.908	1.000	0.954	1.000	1.000	1.000	1.000	1.000	1.000	0.955	1	1.000
Glass5	0.100	1.000	0.400	0.990	1.000	0.937	1.000	0.932	1.000	0.624	1	0.922
Yeast5	0.881	0.903	1.000	0.283	1.000	0.938	1.000	0.956	0.728	0.667	1	0.905
Segment1	0.188	1.000	0.188	1.000	0.992	0.997	0.994	0.997	1.000	0.995	1	0.994
Average	0.694	0.844	0.701	0.776	0.955	0.959	0.969	0.954	0.930	0.876	1	0.900