The Optimization of Feature Selection Based on Chaos Clustering Strategy and Niche Particle Swarm Optimization

Table 2

Experimental results.

Dataset	Method	Ave-size	Error rate (%)
German	All	24	32
	NSGA-II	12.5	31.27
	MOEA/D	8.8	31.09
	CGNPSO	8.58	31
WBCD	All	30	14
	NSGA-II	15	7.2
	MOEA/D	7.6	7.2
	CGNPSO	7.1	7.02
Ionosphere	All	34	17.1
	NSGA-II	15	12
	MOEA/D	10	10.8
	CGNPSO	8.8	10.48
Wine	All	13	37.8
	NSGA-II	8.5	4.04
	MOEA/D	8.0	3.77
	CGNPSO	5.0	3.06
Lung cancer	All	56	30
	NSGA-II	26.5	27.5
	MOEA/D	24	26.7
	CGNPSO	22.01	26.5
Libras	All	90	31.1
	NSGA-II	45	27.8
	MOEA/D	32.7	25.5
	CGNPSO	31.1	25.46
HillValley	All	100	48.5
	NSGA-II	48	42.5
	MOEA/D	37.2	42.7
	CGNPSO	37.05	42.4
Musk1	All	166	16.08
	NSGA-II	84.1	15.12
	MOEA/D	78.9	14.36
	CGNPSO	80.1	14.4
Madelon	All	500	29.1
	NSGA-II	248.9	22.17
	MOEA/D	240.1	22.98
	CGNPSO	239.35	22.88
Isolet5	All	617	1.55
	NSGA-II	309.5	1.51
	MOEA/D	300.2	1.49
	CGNPSO	301.1	1.5