Information Entropy-Based Hybrid Models Improve the Accuracy of Reference Evapotranspiration Forecast

Table 1

Evaluation of forecasting accuracy of single and hybrid forecasting models.

Model types	Model algorithms	Training set	Test set	Training set	Test set	Training set	Test set
		R		RMSE (mm d^−1)		MAPE (%)
Single models	SVR	0.82^b	0.78^b	0.152^b	0.163^b	23.7^c	21.3^b
	Bayesian	0.81^b	0.80^b	0.157^b	0.168^b	24.8^bc	23.8^b
	Lasso	0.79^bc	0.75^bc	0.193^a	0.186^b	30.1^ab	31.7^a
	Ridge	0.75^c	0.71^c	0.216^a	0.233^a	36.9^a	33.6^a
Hybrid models	Information entropy	0.89^a	0.86^a	0.138^c	0.146^c	12.3^d	11.4^c
	Variance reciprocal	0.83^b	0.80^b	0.163^b	0.175^b	20.4^c	19.1^b

Note. R, correlation coefficient; RMSE, root mean square error; MAPE, mean absolute percentage error. Different letters in each column stand for significant differences at .