Determinants of Electricity Consumption of Energy-Vulnerable Group Using Ensemble Gradient-Boosting Algorithm

The increasing energy burden on vulnerable households is critical in modern cities, it crucial to understand how cities can characterize vulnerability and its relationship with the environment. This study modeled relationships between consumption built environmental factors compare determinants average energy-vulnerable households. While conventional approach of identifying often relies household income, this suggested a new by considering group as low-income class high expenditure. A traditional regression model (semi-log regression) advanced machine learning algorithm (ensemble gradient boosting, XGboost) were employed maximize performance modeling processes. results indicated that overall was superior regard algorithm, producing r-squared value 0.92 for households, compared 0.34 semi-log model. direction association similar level exhibited clear difference, especially effect income (comparing 0.30 0.03) housing type -0.45 -0.63). identified several implications regarding urban management policy based findings.