Carbon emission prediction of residential buildings in southern China based on optimised WOA-XGBoost model

To address high carbon emissions (CEs) from residential buildings in southern China, this study proposes an optimised whale optimisation algorithm-eXtreme gradient boosting (WOA-XGBoost) model for CE prediction. The model combines the global search capability of the whale optimisation algorithm with the gradient boosting strategy of XGBoost to overcome the limitations of traditional models. The model's superior performance was verified through experimentation, achieving a mean absolute percentage error (MAPE) as low as 2.56%. It demonstrated high accuracy for typical southern cities, with MAPE values ranging from 8.23% to 9.34%.