. 2020 Jun 29;15(6):e0235324. doi: 10.1371/journal.pone.0235324

Table 12. Performance of Boosting-based models during 10-fold cross validation and testing stages at Guangzhou station.

Models	10 Fold cross validation results			Testing results
	RMSE	MAE	Adj_R²	RMSE	MAE	Adj_R²
	mm/d	mm/d	Adj_R²	mm/d	mm/d	Adj_R²
T_max, T_min, Rs
ADA1	0.4997	0.4115	0.7624	0.5269	0.4457	0.7515
GBDT1	0.3890	0.2610	0.8539	0.3751	0.2614	0.8740
XGB1	0.3778	0.2577	0.8618	0.3620	0.2613	0.8827
LGB1	0.3760	0.2557	0.8629	0.3585	0.2562	0.8849
CAT1	0.3708	0.2516	0.8664	0.3528	0.2533	0.8886
T_max, T_min, RH
ADA2	0.6114	0.5029	0.6453	0.5971	0.4804	0.6808
GBDT2	0.5396	0.4172	0.7228	0.5243	0.3938	0.7539
XGB2	0.5334	0.4186	0.7292	0.5013	0.3846	0.7750
LGB2	0.5253	0.4068	0.7374	0.5005	0.3783	0.7757
CAT2	0.5193	0.4036	0.7432	0.4891	0.3697	0.7858
T_max, T_min, U₂
ADA3	0.7658	0.6215	0.4457	0.7854	0.6315	0.4478
GBDT3	0.6596	0.5088	0.5866	0.6558	0.5006	0.6149
XGB3	0.6422	0.4990	0.6086	0.6341	0.4875	0.6400
LGB3	0.6407	0.4988	0.6106	0.6296	0.4853	0.6451
CAT3	0.6322	0.4906	0.6207	0.6167	0.4760	0.6596
T_max, T_min
ADA4	0.7593	0.6152	0.4558	0.7797	0.6181	0.4561
GBDT4	0.7109	0.5574	0.5210	0.6976	0.5467	0.5645
XGB4	0.6944	0.5501	0.5439	0.6736	0.5365	0.5940
LGB4	0.6872	0.5422	0.5542	0.6654	0.5255	0.6038
CAT4	0.6794	0.5336	0.5625	0.6555	0.5169	0.6156
T_max, T_min, Ra
ADA5	0.7622	0.6171	0.4512	0.7871	0.6389	0.4453
GBDT5	0.6737	0.5221	0.5688	0.6547	0.5084	0.6162
XGB5	0.6592	0.5125	0.5874	0.6427	0.5037	0.6302
LGB5	0.6566	0.5110	0.5904	0.6365	0.4990	0.6373
CAT5	0.6494	0.5036	0.5992	0.6249	0.4879	0.6504
T_max, T_min, J
ADA6	0.7580	0.6138	0.4574	0.7875	0.6386	0.4449
GBDT6	0.6450	0.4946	0.6047	0.6294	0.4838	0.6453
XGB6	0.6312	0.4865	0.6217	0.6127	0.4760	0.6639
LGB6	0.6282	0.4843	0.6252	0.6083	0.4738	0.6687
CAT6	0.6171	0.4745	0.6382	0.5909	0.4560	0.6874