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

Table 3

Recall.

	Dataset	None	SMOTE	ASMOTE	Borderline-SMOTE	SMOTE-RSB*	NRSBoundary-SMOTE
1	Austra	0.8013	0.8404	0.8339	0.8436	0.7948	0.8502
2	Heart-s	0.7333	0.7583	0.7417	0.7667	0.7333	0.7833
3	Bupa	0.5310	0.6483	0.6345	0.6207	0.5310	0.6621
4	Auto-mpg	0.8188	0.9060	0.8993	0.8859	0.8255	0.9195
5	Colic	0.6912	0.7500	0.8088	0.7500	0.6912	0.7574
6	Ionosphere	0.8254	0.8571	0.8730	0.8492	0.8254	0.8730
7	Machine	0.8514	0.8919	0.9054	0.8919	0.8514	0.8919
8	Labor	0.6500	0.5000	0.7500	0.7000	0.7000	0.8000
9	Pima	0.5970	0.7276	0.7388	0.6866	0.5970	0.7313
10	VC	0.6400	0.7700	0.6900	0.7100	0.6400	0.7900
11	German	0.4867	0.4233	0.4267	0.5133	0.4700	0.5500
12	Haberman	0.2963	0.5185	0.5432	0.5432	0.3457	0.5926
13	Transfusion	0.4326	0.4775	0.4438	0.4607	0.3989	0.5000
14	CMC	0.3153	0.4835	0.4535	0.4174	0.3153	0.4985
15	Yeast	0.4754	0.5779	0.5861	0.5123	0.4754	0.5738
	Average	0.6097	0.6754	0.6886	0.6768	0.6130	0.7182

𝑤=0.01 to 0.05.