PMSVM: An Optimized Support Vector Machine Classification Algorithm Based on PCA and Multilevel Grid Search Methods

Table 3

Comparison of the accuracy and efficiency of algorithms on stratified sampled data.

Dataset	Algorithm	Parameters		Time (S)	Accuracy
					Sensitivity	Specificity	Precision	AUC	CVA (%)
S-DATA1	LIBSVM	90.5097	0.011049	19.740543	0.813	0.890	0.856	0.94463	84.4444
	PMSVM	2.82840	0.176780	4.6617020	0.838	0.910	0.882	0.97575	84.4444
S-DATA3	LIBSVM	22.6274	0.062500	20.844411	0.813	0.910	0.878	0.93363	83.8889
	PMSVM	11.3137	0.015625	4.9883590	0.763	0.870	0.824	0.91788	81.6667
S-DATA5	LIBSVM	0.35355	0.125000	20.070109	0.788	0.880	0.840	0.91075	84.4444
	PMSVM	0.50000	0.125000	4.8111860	0.788	0.880	0.840	0.91910	84.4444
S-DATA7	LIBSVM	1.41420	0.353550	20.134157	0.838	0.920	0.893	0.91213	85.0000
	PMSVM	8.00000	0.044194	4.7952370	0.880	0.910	0.877	0.92675	85.0000
S-DATA9	LIBSVM	0.17678	0.083880	19.628079	0.800	0.880	0.842	0.92212	85.5556
	PMSVM	2.82840	0.250000	4.8249560	0.863	0.920	0.896	0.94350	86.6667