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

Table 1

Description of the experimental data sets. Imbalance ratio is the value of 𝑁_maj/𝑁_min.

Data set	Samples	Attributes	Minority class	𝑁min/𝑁maj	Imbalance ratio
Spambase	4601	57	Class 1	1813/2788	1.54
Vote	435	16	Class 1	168/267	1.59
Wdbc	569	30	Malignant	212/357	1.68
Ionosphere	351	33	Bad	126/225	1.79
Pima	768	8	Class 1	268/500	1.87
German	1000	24	Class 2	300/700	2.33
Phoneme	5404	5	Class 1	1586/3818	2.41
Haberman	306	3	Class 2	81/225	2.78
Vehicle	846	18	Opel	212/634	2.99
Cmc	1473	9	Class 2	333/1140	3.42
House	506	13	[20, 21]	106/400	3.77
Scrapie	3113	14	Class 1	531/2582	4.86
Yeast	1484	5	Class 4	163/1321	8.10
Mfeat_zer	2000	47	Digit 9	200/1800	9.00
Mfeat_kar	2000	64	Digit 9	200/1800	9.00
Satimage	6435	36	Class 4	626/5809	9.28
Abalone7	4177	8	Class 7	391/3786	9.68
Sick	3163	25	Class 1	293/2870	9.80
Cbands	12000	30	Class 1	500/11500	23.00
Ozone	2536	72	Class 1	73/2463	33.74