A Novel Time Series Prediction Approach Based on a Hybridization of Least Squares Support Vector Regression and Swarm Intelligence

Table 3

Result comparison.

Time series	Result comparison		BPNN	ANFIS	ESIM	FLSVRTSP
Mackey-Glass	Training	RMSE	0.029	0.013	0.073	0.005
		MAE	0.024	0.008	0.059	0.003
	Testing	RMSE	0.027	0.011	0.069	0.005
		MAE	0.023	0.007	0.057	0.004
Water flow	Training	RMSE	26.66	20.66	25.34	25.15
		MAE	21.18	15.69	19.31	19.08
	Testing	RMSE	15.33	14.13	10.71	10.33
		MAE	12.55	11.30	8.88	8.35
Exchange rate	Training	RMSE	0.42	0.20	0.44	0.39
		MAE	0.31	0.13	0.32	0.26
	Testing	RMSE	0.60	0.67	0.42	0.36
		MAE	0.47	0.52	0.33	0.29