A Forecasting Approach Combining Self-Organizing Map with Support Vector Regression for Reservoir Inflow during Typhoon Periods

Table 3

Performance comparison of the SVM, SOM-SVM1, and SOM-SVM2 models for 1 h ahead forecasts using RRMSE and CE as criteria.

Typhoon events	SOM		SOM-SVM1		SOM-SVM2
	RRMSE	CE	RRMSE	CE	RRMSE	CE
Polly	1.358	0.583	0.448	0.912	0.374	0.916
Ted	0.316	0.770	0.157	0.941	0.128	0.958
Tim	3.627	0.551	1.724	0.873	1.253	0.901
Fred	1.460	0.745	0.369	0.960	0.365	0.952
Gladys	7.430	0.644	1.354	0.921	0.545	0.917
Seth	5.882	0.650	0.817	0.936	0.731	0.950
Herb	1.375	0.959	0.367	0.967	0.268	0.975
Zeb	0.340	0.905	0.187	0.958	0.186	0.955
Xangsane	0.280	0.918	0.187	0.915	0.151	0.936
Nari	0.268	0.808	0.263	0.895	0.222	0.925
Haiyan	1.222	0.743	0.471	0.912	0.337	0.940
Rammasun	2.975	0.643	0.740	0.929	0.613	0.929
Aere	2.784	0.935	1.245	0.979	1.057	0.988
Nock-Ten	0.919	0.788	0.769	0.961	0.657	0.937
Nanmadol	1.681	0.806	0.258	0.924	0.237	0.908
Haitang	7.049	0.727	2.329	0.923	1.225	0.948
Matsa	4.292	0.523	1.474	0.904	1.032	0.900
Talim	3.623	0.874	0.874	0.883	0.779	0.958
Krosa	0.667	0.960	0.554	0.980	0.402	0.981
Fung-Wong	0.876	0.830	0.286	0.979	0.262	0.980
Sinlaku	0.696	0.858	0.212	0.939	0.226	0.913
Jangmi	0.400	0.937	0.133	0.986	0.132	0.978