Bon Seo, Sangho Choi, ByungYun Lee, Sojin Lee, Su-Gwang Jeong
Buildings are a major source of urban greenhouse gas emissions, yet district-level heterogeneity in building energy inefficiency remains insufficiently incorporated into environmental management and decarbonization planning. Evidence-based intervention prioritization is particularly important in high-density cities, where building stocks, urban form, and seasonal demand vary across districts. Using Seoul as an empirical case, 2022 electricity and gas consumption data were analyzed to identify key factors associated with energy use intensity (EUI) and support differentiated district-level intervention prioritization. A total of 115,960 valid electricity and 78,723 gas records were obtained after filtering for missing and extreme values. K-means clustering classified the 25 districts into four groups based on annual mean electricity and gas EUIs. The regression models explained 97.64% and 96.44% of the variation in electricity and gas EUI, respectively. Summer and winter peak indicators showed the strongest associations with electricity and gas EUI, respectively. The models revealed distinct electricity-gas asymmetries across building characteristics. Building on the regression results, a residual-based assessment of district energy performance gaps identified seven dual-high districts as priority intervention targets, and Moran's I and Local Moran's I analyses confirmed spatially clustered residual patterns, including high-high clusters in northern and northeastern Seoul. Scenario-based GHG estimation further revealed that a 10% EUI reduction in the seven dual-high districts would yield 577 ktCO2, the highest reduction potential among all residual-based groups, indicating their high priority for district-level decarbonization. The proposed framework provides a practical basis for identifying districts with higher-than-expected energy use and supporting spatially differentiated building-stock decarbonization in high-density cities.