An Empirical Study on the Performance of Cost-Sensitive Boosting Algorithms with Different Levels of Class Imbalance

Table 4

Index of performance using 𝐺-mean for severely imbalanced data sets.

	Win	Loss	Index	Win	Loss	Index	Win	Loss	Index	Win	Loss	Index	Win	Loss	Index
	0.1			0.2			0.3			0.4			0.5
AdaC1	7	0	7	9	0	9	8	0	8	10	0	10	12	0	12
AdaC2	23	1	22	31	1	30	30	1	29	30	1	29	28	1	27
AdaC3	13	5	8	16	8	8	18	8	10	14	8	6	14	8	6
AdaCost	1	15	−14	0	21	−21	0	20	−20	0	22	−22	0	25	−25
CSB1	1	15	−14	1	15	−14	1	16	−15	1	14	−13	1	12	−11
CSB2	0	9	−9	0	12	−12	0	12	−12	0	10	−10	0	9	−9
	0.6			0.7			0.8			0.9			1
AdaC1	12	0	12	14	0	14	15	0	15	15	0	15	15	0	15
AdaC2	24	1	23	26	1	25	24	1	23	17	1	16	15	0	15
AdaC3	14	5	9	13	7	6	13	5	8	17	2	15	15	0	15
AdaCost	0	26	−26	0	27	−27	0	27	−27	0	27	−27	0	27	−27
CSB1	1	12	−11	0	14	−14	0	15	−15	0	16	−16	0	18	−18
CSB2	0	7	−7	0	4	−4	0	4	−4	0	3	−3	0	0	0