Constructing Better Classifier Ensemble Based on Weighted Accuracy and Diversity Measure

Table 3

The average size of ensembles selected by candidate measures and GA.

Dataset	WAD	Kappa-Error	GenDiv	Acc-Only	Div-Only	No-Selection
Hepatitis	23	11	40	10	25	200
Autos	20	22	34	9	32	200
Heart-statlog	28	13	46	8	34	200
Ionosphere	30	10	34	13	45	200
Credit-a	25	5	34	15	56	200
Diabetes	27	12	34	12	125	200
Vehicle	22	3	27	14	111	200
Car	23	23	32	7	78	200
cmc	26	17	89	6	65	200
Segment	25	19	78	1	34	200
kr-vs-kp	20	16	45	6	43	200
Hypothyroid	29	2	78	18	76	200
Waveform	19	9	10	16	56	200
Page-blocks	24	20	34	11	12	200
Nursery	23	7	65	13	34	200
Average	24.3	12.6	45.3	10.6	55.1	200