A Parallel Genetic Algorithm Based Feature Selection and Parameter Optimization for Support Vector Machine

Table 4

Experiment results for Australian dataset using our proposed approach and grid search.

Fold number	Our proposed approach					Grid search method
			Number of features	Number of support vectors	Accuracy (%)			Number of features	Number of support vectors	Accuracy
1	368.08299	1.48664	5	173	85.5072	32	0.125	14	202	82.6087
2	827.07180	1.65084	6	206	91.3043	0.5	2.0	14	324	89.8551
3	529.60439	0.32159	7	203	89.8551	0.5	1.0	14	261	86.7647
4	1845.7247	0.95494	3	198	94.2857	0.125	0.03125	14	499	81.1594
5	115.07262	4.63776	6	215	92.8571	512	0.0625	14	193	86.9565
6	347.31311	0.29031	7	197	89.8551	1024	0.03125	14	197	89.8551
7	379.01391	6.52218	3	214	89.7059	0.5	2.0	14	319	88.2353
8	131.80564	4.06696	5	204	90.0000	128	0.125	14	189	76.8116
9	584.01293	5.78718	3	229	88.2353	32	0.125	14	205	82.8571
10	74.464588	3.41797	6	188	89.7059	0.125	0.015625	14	504	90.0000
Avg			5.1	202.7	90.1312			14	289.3	85.5104