Carbon Trading Price Prediction Based on Complex Wavelet Transform and Gaussian Process Twice Decompose

With the advancement of China's carbon trading market, accurate price prediction is essential for market stability. This study proposes a dual decomposition strategy using Complex Wavelet Transform and Gaussian Process Regression, with Kolmogorov complexity and log marginal likelihood for layer determination, and Approximate Entropy for sesquence reconstruction. A Gated Recurrent Unit serves as the prediction model, optimized via the CPO Algorithm, achieves high accuracy in predicting carbon prices in Wuhan and Beijing.