Exploring an Interactive Value-Adding Data-Driven Model of Consumer Electronics Supply Chain Based on Least Squares Support Vector Machine

Table 4

Comparison of the accuracies of the LS-SVM models based on different kernel functions.

Samples	Primary indicator			Secondary indicator			Tertiary indicator
	RBF	LINEAR	POLY	RBF	LINEAR	POLY	RBF	LINEAR	POLY
Training set	0.000	0.018	0.003	0.021	0.021	0.040	0.002	0.003	0.003
	0.000	0.031	0.036	0.044	0.054	0.100	0.017	0.014	0.024
	0.000	0.012	0.015	0.032	0.022	0.045	0.012	0.008	0.015
	0.000	0.020	0.029	0.071	0.096	0.093	0.005	0.005	0.009
	0.000	0.118	0.186	0.093	0.094	0.188	0.022	0.022	0.041
	0.000	0.078	0.064	0.124	0.135	0.102	0.022	0.012	0.022
	0.000	0.234	0.157	0.145	0.050	0.127	0.017	0.006	0.010
Test set	0.053	0.057	0.024	0.059	0.085	0.082	0.071	0.072	0.076
	0.050	0.011	0.018	0.033	0.022	0.016	0.001	0.044	0.044
	0.010	0.063	0.091	0.069	0.190	0.134	0.001	0.062	0.075
Mean error (test set)	0.038	0.044	0.044	0.054	0.099	0.077	0.024	0.059	0.065
Mean error (all sets)	0.011	0.064	0.062	0.071	0.077	0.093	0.017	0.025	0.032