A Novel Boundary Oversampling Algorithm Based on Neighborhood Rough Set Model: NRSBoundary-SMOTE

Table 5

AUC.

	Dataset	None	SMOTE	ASMOTE	Borderline-SMOTE	SMOTE-RSB*	NRSBoundary-SMOTE
1	Austra	0.8483	0.8739	0.8433	0.8635	0.8458	0.8657
2	Heart-s	0.7443	0.7947	0.7223	0.7458	0.7448	0.8146
3	Bupa	0.6650	0.6468	0.6132	0.6355	0.6652	0.6401
4	Auto-mpg	0.9282	0.8973	0.8874	0.9015	0.9303	0.9197
5	Colic	0.7873	0.7740	0.8170	0.7971	0.7855	0.8102
6	Ionosphere	0.8923	0.8879	0.9130	0.8778	0.8930	0.9078
7	Machine	0.9359	0.9199	0.9268	0.9538	0.9359	0.9430
8	Labor	0.7588	0.7500	0.8047	0.7655	0.7655	0.8243
9	Pima	0.7514	0.7417	0.7321	0.7158	0.7513	0.7419
10	VC	0.8380	0.8107	0.7688	0.8044	0.8380	0.8277
11	German	0.6884	0.6333	0.6115	0.6853	0.6620	0.6791
12	Haberman	0.6087	0.6255	0.6570	0.6536	0.6174	0.6636
13	Transfusion	0.7001	0.6813	0.6836	0.7075	0.7007	0.7048
14	CMC	0.6373	0.6866	0.6723	0.6710	0.6375	0.6980
15	Yeast	0.7500	0.7801	0.7698	0.7649	0.7500	0.7831
	Average	0.7689	0.7669	0.7615	0.7695	0.7682	0.7882

𝑤=0.01 to 0.05.