A Novel Selective Ensemble Algorithm for Imbalanced Data Classification Based on Exploratory Undersampling

Table 9

Pairwise comparisons of all algorithms based on the 𝐹-measure criterion.

Algorithms		CART	RUSB	SMOB	UNBag	SMBag	AdaC	RAMO	RotF	Easy	RotE-un	RotEasy
Mean		0.5991	0.6728	0.6890	0.6522	0.6494	0.5917	0.6704	0.6365	0.6769	0.6776	0.7072
CART	𝑟̇		1.1397	1.1664	1.1304	1.1010	0.9199	1.1149	0.9823	1.1670	1.1664	1.2320
	𝑠		17-0-3	19-0-1	17-0-3	19-0-1	12-0-8	18-0-2	15-0-5	19-0-1	19-0-1	19-0-1
	𝑞		0.0015	0.0000	0.0015	0.0000	0.3833	0.0002	0.0266	0.0000	0.0000	0.0000
RUSB	𝑟̇			1.0234	0.9918	0.9661	0.8072	0.9783	0.8619	1.0239	1.0235	1.0810
	𝑠			13-0-7	9-0-11	7-0-13	6-0-14	10-0-10	5-0-15	11-0-9	11-0-9	14-0-6
	𝑞			0.1892	0.6636	0.1892	0.0784	1.0000	0.0266	0.6636	0.6636	0.0784
SMOB	𝑟̇				0.9691	0.9440	0.7887	0.9559	0.8422	1.0005	1.0001	1.0563
	𝑠				8-0-12	3-0-17	2-0-18	4-0-16	3-0-17	10-0-10	7-0-13	13-0-7
	𝑞				0.3833	0.0015	0.0002	0.0072	0.0015	1.0000	0.1892	0.1892
UNBag	𝑟̇					0.9740	0.8138	0.9863	0.8690	1.0324	1.0319	1.0899
	𝑠					12-0-8	10-0-10	10-0-10	10-0-10	16-0-4	13-0-7	20-0-0
	𝑞					0.3833	1.0000	1.0000	1.0000	0.0072	0.1892	0.0000
SMBag	𝑟̇						0.8355	1.0126	0.8922	1.0599	1.0594	1.1190
	𝑠						10-0-10	15-0-5	12-0-8	13-0-7	13-0-7	14-0-6
	𝑞						1.0000	0.0266	0.3833	0.1892	0.1892	0.0784
AdaC	𝑟̇							1.2120	1.0678	1.2685	1.2679	1.3393
	𝑠							17-0-3	12-0-8	14-0-6	14-0-6	14-0-6
	𝑞							0.0015	0.3833	0.0784	0.0784	0.0784
RAMO	𝑟̇								0.8810	1.0467	1.0462	1.1050
	𝑠								4-0-16	11-0-9	10-0-10	13-0-7
	𝑞								0.0072	0.6636	1.0000	0.1892
RotF	𝑟̇									1.1880	1.1875	1.2542
	𝑠									13-0-7	12-0-8	15-0-5
	𝑞									0.1892	0.3833	0.0266
Easy	𝑟̇										0.9996	1.0558
	𝑠										11-0-9	14-0-6
	𝑞										0.6636	0.0784
RotE-un	𝑟̇											1.0562
	𝑠											16-0-4
	𝑞											0.0072