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

Table 5

Comparison of the LHLGSA, LMARS, LM5RT, and LCMLR methods: Astore catchment.

Indexes	Cross validation data sets	Input combinations	Methods
			LHLGSA	LMARS	LM5RT	LCMLR
	DS1	IC1	54.73	62.76	70.22	123.24
		IC2	43.31	50.88	69.52	65.05
		IC3	52.54	54.65	71.86	60.76
	DS2	IC1	72.25	78.42	80.68	135.75
		IC2	56.78	59.43	74.33	65.48
		IC3	58.11	64.52	83.89	68.60
	DS3	IC1	56.65	72.06	64.28	129.02
		IC2	45.75	48.49	57.39	61.24
		IC3	48.69	53.97	64.91	54.83
	DS4	IC1	43.36	59.96	61.91	96.04
		IC2	35.33	38.56	46.18	40.11
		IC3	36.72	38.78	51.49	39.02
	DS1	IC1	33.28	43.89	40.52	60.16
		IC2	26.91	27.05	38.73	31.79
		IC3	28.51	29.98	40.32	30.29
	DS2	IC1	42.04	54.60	47.71	66.67
		IC2	29.12	30.29	41.03	32.68
		IC3	30.62	36.75	45.64	31.15
	DS3	IC1	32.83	38.91	34.88	60.72
		IC2	27.27	26.81	28.27	27.54
		IC3	29.37	28.06	32.86	27.07
	DS4	IC1	20.29	39.92	34.89	52.10
		IC2	13.08	20.88	24.80	21.42
		IC3	17.58	21.06	27.02	21.08
	DS1	IC1	0.842	0.806	0.741	0.714
		IC2	0.894	0.877	0.789	0.824
		IC3	0.882	0.859	0.763	0.846
	DS2	IC1	0.822	0.785	0.760	0.695
		IC2	0.882	0.876	0.775	0.820
		IC3	0.866	0.845	0.729	0.836
	DS3	IC1	0.856	0.798	0.815	0.736
		IC2	0.905	0.890	0.817	0.836
		IC3	0.896	0.893	0.779	0.855
	DS4	IC1	0.894	0.854	0.843	0.746
		IC2	0.921	0.906	0.899	0.896
		IC3	0.914	0.896	0.888	0.901