Fuzzy Weighted Least Squares Support Vector Regression with Data Reduction for Nonlinear System Modeling

Table 1

Comparison results of the proposed method, [15], G-LSSVR and Local-LSSVR with the hyper-parameter set , and the overlap factor as 1.5 are shown for Example 1.

: the number of the SRMs	RMSE (Training)		RMSE (Testing)
	our approach	[15]	our approach	[15]
3-SRMs	0.2172	0.8184	0.2151	0.8186
5-SRMS	0.0368	0.5446	0.0363	0.5447
7-SRMs	0.0146	0.3554	0.0140	0.3556
9-SRMs	0.0079	0.3389	0.0076	0.3391
12-SRMS	0.0042	0.1957	0.0040	0.1961
15-SRMs	0.0026	0.1471	0.0026	0.1472
G-LSSVR	1.7295		1.7298
Local-LSSVR
M=21 training points	3.7635		3.7630
M=41 training points	1.1742		1.1622
M=61 training points	0.0054		0.0050
M=81 training points	0.0028		0.0026