Effective Estimation of Hourly Global Solar Radiation Using Machine Learning Algorithms

Table 9

Estimation results of the most successful models that were developed according to LibSVM and the input data of both provinces.

Target provinces		Kahramanmaras				Isparta
Data groups	SVM type	RMSE	MAE	SMAPE (%)		RMSE	MAE	SMAPE (%)
GR1	E-SVR	0.1271	0.1003	15.11	0.7876	0.1431	0.1153	15.46	0.6372
	Nu-SVR	0.1257	0.0960	14.97	0.7893	0.1428	0.1144	15.44	0.6377
GR2	E-SVR	0.0742	0.0584	13.43	0.9278	0.0816	0.0637	12.64	0.8818
	Nu-SVR	0.0675	0.0501	12.14	0.9394	0.0765	0.0580	12.05	0.8961
GR3	E-SVR	0.0765	0.0604	13.69	0.9229	0.0815	0.0643	12.95	0.8821
	Nu-SVR	0.0697	0.0523	12.42	0.9352	0.0752	0.0573	12.11	0.8995
GR4	E-SVR	0.1300	0.1026	15.17	0.7781	0.1495	0.1193	15.66	0.6050
	Nu-SVR	0.1287	0.0984	14.97	0.7792	0.1491	0.1185	15.64	0.6057
GR5	E-SVR	0.0780	0.0616	13.77	0.9197	0.0849	0.0673	12.28	0.8728
	Nu-SVR	0.0717	0.0540	12.59	0.9315	0.0827	0.0621	11.77	0.8792
GR6	E-SVR	0.0802	0.0631	13.80	0.9146	0.0827	0.0645	12.52	0.8790
	Nu-SVR	0.0756	0.0576	12.96	0.9237	0.0779	0.0594	12.15	0.8925