Predicting and decarbonizing carbon emissions from building energy use in Hong Kong: A LASSO regression approach

Abstract

This study examines the total carbon emissions of Hong Kong, which ranked 74th among 218 countries in 2021, with a focus on the contribution of building energy use. Using data from open sources at ourworldindata.org and data.gov.hk from 2001 to 2020, a Least Absolute Shrinkage and Selection Operator (LASSO) regression model is developed to predict the trajectory of carbon emissions. The significance of regression coefficients is tested using bootstrap confidence intervals, and jackknife cross-validation is employed to evaluate the model's performance. The study identifies nine significant predictors, including commercial electricity consumption, private office area, and private commercial building area, among others. The findings indicate that the current carbon reduction trend is insufficient to meet the target in 2030. To achieve carbon neutrality by 2050, the study proposes a decarbonization roadmap involving a 30–40 % reduction in commercial building electricity consumption and a decreased carbon emission factor. The study underscores the importance of strategic energy management for air conditioning systems and the promotion of green building projects in effectively decarbonizing Hong Kong.