Forecasting European Carbon Price Based on CEEMDAN-GRU Hybrid Multi-scale Model
Based on the nonlinear,non-stationary and multi-scale characteristics of carbon prices,a CEEMDAN-GRU hybrid forecasting model was constructed to perform nonlinear prediction of carbon price,and the out-of-sample term heterogeneity prediction was used to examine its robustness.The results showed that compared with the hybrid models constructed by EMD,CEEMD technologies,and LSTM-BP models,the CEEMDAN-GRU model can effectively capture the multi-scale time-frequency characteristics of carbon prices,with the forecasting errors of RMSE,MAE and MAPE reaching 1.021 8,0.681 5 and 0.011 0 respectively.The carbon price fore-casting accuracy was superior to the benchmark models,especially the short-term forecasting performance.
NETL
NSTL
万方数据
