Improving Accuracy of River Flow Forecasting Using LSSVR with Gravitational Search Algorithm

Table 4

Comparison of HLGSA performance for different kernel functions using best data set of Astore and Shyok catchments.

Catchment	Performance indexes	Kernel functions
		RBF	Polynomial	Sigmoid	Gaussian	Morlet	Mexican hat	Mayer
Astore		40.09	70.55	67.12	66.88	66.51	65.57	57.90
		25.05	42.25	41.48	39.55	39.53	39.38	34.27
		0.916	0.878	0.892	0.889	0.887	0.901	0.903
Shyok		109.20	156.53	155.45	138.47	137.72	137.95	144.92
		55.75	83.72	82.40	72.52	70.93	76.30	83.21
		0.947	0.873	0.875	0.901	0.901	0.908	0.903