Carbon Emission Right Price Prediction based on IOWHA Operator Vector Angle Cosine Combination Deep Learning

Abstract

The relentless exacerbation of the greenhouse effect poses a significant challenge to the progress and sustainability of human civilization. Carbon emission rights serve as a pivotal gauge of the carbon trading market's dynamics. Examining the impact of carbon emission rights prices can shed light on how governments indirectly influence these prices through mechanisms such as carbon emission quotas and permits. This study focuses on analyzing the transaction prices of carbon emission rights quotas within the Guangzhou carbon trading market spanning from January 2015 to March 2023. Leveraging a selection of 17 variables across the realms of energy prices, macroeconomics, financial markets, and ecological environment, and drawing insights from the predictive outcomes of three distinct sub-models—namely Ridge Regression, Random Forest Regression, and Long Short-Term Memory neural network (LSTM)—a novel IOWHA operator vector angle cosine combination model is devised. The findings underscore that this combined model harnesses the predictive strengths of each sub-model, effectively curbing prediction discrepancies, and enhancing overall accuracy. The integration of these methodologies represents an optimal fusion for predictive modeling.