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

Table 1

Data description.

	Datasets	Size	Attribute	Class label (minority : majority)	Class distribution
1	Austra	690	14C	1 : 0	307/383
2	Heart-s	270	6C 7N	Present : absent	120/150
3	Bupa	345	6C	1 : 2	145/200
4	Auto-mpg	398	5C 2N	Others : 1	149/249
5	Colic	368	7C 15N	No : yes	136/232
6	Ionosphere	351	34C	b : g	126/225
7	Machine	209	7C	Others : 2	74/135
8	Labor	57	8C 8N	Bad : good	20/37
9	Pima	768	8C	1 : 0	268/500
10	Vertebral column (VC)	310	7C	Normal : abnormal	100/210
11	German	1000	24C	2 : 1	300/700
12	Haberman	306	3C	2 : 1	81/225
13	Transfusion	748	4C	1 : 0	178/570
14	Contraceptive method choice (CMC)	1473	9C	2 : others	333/1140
15	Yeast	1484	8C	MIT : others	244/1240

C: continuous, N: nominal.