Boosting methods for multi-class imbalanced data classification: an experimental review

Journal of Big Data

Table 9 Ranks computed by the Wilcoxon test based on the MMCCs

Algorithms	AdaBoost.MH	SAMME	CatBoost	LogitBoost	GradientBoost	XGBoost	MEBoost	SMOTEBoost	RUSBoost	LightGBM	AdaC1	AdaC2	AdaC3	AdaCost
AdaBoost.MH	−	63	16	36	62	25	82	45	84	82	76	75	61	59
SAMME	57	−	12	30	55	43	62	29	90	74	93	87	70	64
CatBoost	104	108	−	82	95	84	105	89	116	99	110	110	107	104
LogitBoost	84	90	38	−	73	58	84	79	96	85	104	106	89	89
GradientBoost	58	65	25	47	−	50	62	50	75	66	79	71	74	59
XGBoost	95	77	36	62	70	−	75	61	95	83	106	104	85	83
MEBoost	38	58	15	36	58	30	−	43	75	83	75	65	59	57
SMOTEBoost	75	91	31	41	70	59	77	−	104	82	109	113	95	89
RUSBoost	36	30	4	24	45	25	45	16	−	68	68	54	39	48
LightGBM	38	46	21	35	54	37	37	38	52	−	63	49	44	50
AdaC1	44	27	10	16	41	14	45	11	52	57	−	32	31	35
AdaC2	45	33	10	14	49	16	55	7	66	71	88	−	35	50
AdaC3	59	50	13	31	46	35	61	25	81	76	89	85	−	59
AdaCost	61	56	16	31	61	37	63	31	72	70	85	70	61	−