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

Table 6

Index of performance using 𝐹-measure for data sets with a low-level imbalance.

	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	16	0	16	10	0	10	7	0	7	7	0	7	5	0	5
AdaC2	5	5	0	2	4	−2	2	3	−1	9	2	7	11	2	9
AdaC3	1	9	−8	1	5	−4	3	2	1	12	1	11	15	0	15
AdaCost	5	2	3	2	2	0	2	4	−2	3	7	−4	1	11	−10
CSB1	3	3	0	2	2	0	2	3	−1	2	12	−10	0	12	−12
CSB2	0	11	−11	0	4	−4	0	4	−4	0	11	−11	0	7	−7
	0.6			0.7			0.8			0.9			1
AdaC1	10	0	10	9	0	9	10	0	10	11	0	11	10	0	10
AdaC2	15	1	14	16	1	15	17	2	15	13	0	13	13	0	13
AdaC3	18	0	18	20	0	20	17	0	17	13	0	13	10	0	10
AdaCost	0	20	−20	0	22	−22	1	23	−22	0	21	−21	1	21	−20
CSB1	0	12	−12	1	11	−10	0	14	−14	0	12	−12	1	11	−10
CSB2	0	10	−10	0	12	−12	0	6	−6	0	4	−4	0	3	−3