A Novel Hybrid Deep Learning Model for Sugar Price Forecasting Based on Time Series Decomposition

Table 6

Comparison of the predictive accuracy of the various models.

Models	MAE	MAPE	RMSE
ARIMA	4.135	1.189	5.747
SVR	30.877	9.251	33.884
DNN	12.679	3.675	15.915
EMD-DNN	14.083	3.968	14.756
TPE-EMD-DNN	2.648	0.748	3.193