Effective Estimation of Hourly Global Solar Radiation Using Machine Learning Algorithms

Table 13

Comparison of the performance of the present study with various HAGSR estimation models of the previous studies.

ML methods	Best methods	Time range	Location	Author [Ref.]	Statistical indicators
					RMSE	MAE
ANN	ANN	2001–2007	La Serena (Chile)	Lazzús et al. [18]	—	—	0.9437
ANN	ANN	02 February–31 May 2011	Algeria	Hasni et al. [17]	0.172	2.9971	0.9999
MLP	MLP	2009–2012	Fez (Morocco)	Ihya et al. [62]	—	—	0.8896
MLP-NARX	NARX	2010–2014	Fes (Morocco)	Loutfi et al. [32]	—	—	0.95
AdaBoost, LR, KNN, CART, SVR, ANN, RD regression	RD regression	2013–2015	South Korea	Kim et al. [63]	577.5	—	0.705
ANN	ANN	2006–2010	Ajaccio (Corsica)	Notton et al. [12]	12.43	19.17	0.998
MARS-ANN-LR	ANN	2010–2016	Hong Kong	Li et al. [28]	0.270	0.194	0.918
MFFNN-KNN-M5 rules-LibSVM	MFFNN	2002–2006	Kahramanmaras-Isparta (Turkey)	Present study	0.0341	0.0508	0.9656