Analysis and prediction of carbon emission factors based on genetic programming

• GP demonstrates superior predictive accuracy compared to OLS • GP is its random nature. • The GP model can be implemented through multiple approaches. In recent years, frequent extreme weather events such as storms and floods globally have made carbon emission reduction a critical issue for addressing environmental challenges through sustainable green pathways. This study employs least squares regression and genetic programming symbolic regression techniques based on carbon emission data from 2013 to 2023 to investigate factors influencing carbon dioxide emission intensity, total emissions, and their respective weights. The study further compares the predictive capabilities of the least squares regression model with three genetic programming symbolic regression models. Results indicate that the significance and weight values of variables in the least squares regression model exhibit similarities with those in the genetic programming symbolic regression models. All three genetic programming symbolic regression models demonstrate superior predictive capabilities compared to the least squares regression model.