Carbon price decomposition ensemble hybrid forecasting model based multi-scale feature extraction

Abstract

The carbon emission market is the core policy tool to achieve the goal of carbon peaking and carbon neutrality. To fully extract the complex features of carbon price series such as non-stationary, non-linear, and multi-scale etc. This paper constructs an integrated hybrid forecasting model CEEMD-GWO-LSSVR based on the multi-scale decomposition of carbon price decomposition. Firstly, the original carbon price series are decomposed into eigenmodal functions (IMFs) of different scales by complementary ensemble empirical modal decomposition (CEEMD), and the LSSVR model optimized by the grey wolf optimization algorithm (GWO) is used as the prediction model to forecast the obtained IMFs, and finally, the prediction results of all IMFs are linearly integrated. This paper selects the price data of the Shanghai carbon trading market for the empirical study, and the empirical results show that the prediction accuracy of the hybrid model proposed in this paper is significantly better than that of the benchmark model.