Analysis of determinants of building energy use in Seoul through clustering and regression, and proposal of tailored energy policies

Department of Architectural Engineering, Soongsil University, Seoul 06978, Republic of Korea

^* Corresponding author: This email address is being protected from spambots. You need JavaScript enabled to view it.

Abstract

Buildings account for a major share of greenhouse gas emissions, particularly in high-density cities such as Seoul, where over two-thirds of energy-related emissions come from the building sector. Globally, citylevel mitigation strategies are emphasized to achieve carbon neutrality (Net Zero), with improvements in building energy efficiency recognized as one of the most effective measures. This study analyzes electricity and city gas data of Seoul's buildings to identify the determinants of energy use intensity (EUI) and propose reduction strategies tailored to district characteristics. Electricity and gas data from 2022, including monthly consumption, building type, floor area, and district information, were preprocessed, yielding 115,960 electricity and 78,723 gas records. Based on the annual mean electricity and gas EUI of 25 districts, K-means clustering classified the districts into four groups. Multiple linear regression was then conducted for electricity and gas EUI in each group using numerical variables (SVF, summer peak, winter peak, summer minimum, log(area)) and categorical variables (building types with more than 1,000 records). Results demonstrated that electricity EUI was most strongly influenced by summer peak (standardized β=+0.57, p<0.01), while gas EUI was dominated by winter peak (β=+0.76, p<0.01), underscoring the significance of seasonal loads. Among categorical variables, multi-family residential buildings in Group D showed the largest positive coefficient for gas EUI (B=+30.96), indicating a strong dependency on heating and hot water, whereas elderly and childcare facilities in Group C exhibited notable gas-reducing effects (B=-9.86). Accommodation facilities consistently displayed significant positive coefficients across all groups for both electricity (B=+1.90) and gas (B=+9.90), with Group B recording the highest coefficients (electricity: B=+2.55, gas: B=+17.01). These findings highlight the limitations of uniform policy approaches and suggest that tailored strategies—such as high-efficiency retrofits for accommodation facilities, heating upgrades in multi-family residential buildings, and dissemination of best practices from efficient facilities—are essential for improving building energy performance and reducing emissions.