A multifactor hybrid model for carbon price interval prediction based on decomposition-integration framework.

Area of Science:

• Environmental Economics
• Computational Finance
• Data Science

Background:

• Accurate carbon price estimation is crucial for effective carbon trading and emission reduction policies.
• Existing prediction models often lack accuracy or comprehensive interval prediction capabilities.
• Volatility and influencing factors in carbon markets necessitate advanced modeling techniques.

Purpose of the Study:

• To develop a novel hybrid prediction model for carbon price, incorporating both point and interval estimations.
• To enhance the accuracy and reliability of carbon price forecasts for market participants.
• To improve the operational efficiency of carbon markets and support emission reduction strategies.

Main Methods:

• Adaptive decomposition of carbon price using successive variational mode decomposition.
• Feature selection via partial autocorrelation function and random forest for optimal input variables.
• Hybrid point prediction using categorical boosting and kernel extreme learning machine, optimized by sparrow search algorithm.
• Interval prediction utilizing adaptive bandwidth kernel density estimation.
• Model interpretability through Shapley additive explanation.

Main Results:

• The proposed hybrid model achieved high accuracy with MAE of 0.1022, MAPE of 0.0022, RMSE of 0.1262, and R² of 0.9921 on Hubei carbon market data.
• Historical carbon price, Brent crude oil futures, and EU allowance futures positively impacted carbon price; Hushen 300 showed a negative impact.
• The model demonstrated superior interval prediction performance with higher coverage probability and narrower interval width compared to constant kernel density estimation.

Conclusions:

• The developed hybrid model offers a significant advancement in carbon price prediction accuracy and reliability.
• The model's ability to provide both accurate point and interval predictions aids market participants and policy implementation.
• This approach can enhance carbon market operations and contribute to achieving climate change mitigation goals.