Forecasting China carbon price using a error corrected secondary decomposition hybrid model integrated fuzzy dispersion entropy and deep learning paradigm

Abstract

Abstract To promote the market-oriented mechanism in carbon emission reduction, improve the role of carbon price forecasting in guiding investors to make quantitative investments, this paper constructs a error corrected carbon price forecasting model integrated fuzzy dispersion entropy and deep learning paradigm, namely, ICEEMDAN-FDE-VMD-PSO-LSTM-EC. Initially, the ICEEDMAN is used to primary decompose the original carbon price. Subsequently, the fuzzy dispersion entropy is conducted to identify the high-complexity signal after the primary decomposition. Thirdly, the VMD and deep learning paradigm of LSTM optimized by the PSO algorithm are employed to secondary decompose the high complexity signals and perform the out-of-sample forecasting. Finally, the error corrected (EC) method is conducted to re-modify the above predicted results to improve the forecasting accuracy. The results conclude that the forecasting performance of the ICEEMDAN-type secondary decomposition models are significantly better than the primary decomposition models, the deep learning PSO-LSTM-type models have superiority in forecasting China carbon price, the error corrected method for improving the forecasting accuracy has achieve satisfactory results. Noteworthy, the proposed model has the best forecasting accuracy, with the forecasting errors RMSE, MAE, RMSE and Pearson correction are 0.0877, 0.0407, 0.0009 and 0.9998. Especially, the long-term forecasting performance for 750 consecutive trading price is outstanding. Those conclusions contribute to judge the carbon price characteristics and formulate market regulations.