Data-driven characterization of electricity demand regimes in a university campus served by geothermal district heating

University campuses are complex energy systems in which electricity demand reflects the combined influence of weather, occupancy schedules, base loads, and auxiliary infrastructure. In campuses served by geothermal district heating, electricity consumption does not measure heating demand directly, yet it is typically the only continuously monitored energy signal available to campus operators. This study characterizes daily electricity demand regimes at Afyon Kocatepe University, Türkiye, using fourteen months of 15-min electricity and on-site meteorological measurements. Heating degree days showed a positive but limited association with daily consumption ( r = 0.393, R² = 0.155; slope 145.1 kWh per °C·day, 95% CI 105.2–185.0 under autocorrelation-robust inference), and this association was insensitive to the assumed base temperature (15–19 °C). K-Means separated a cold-season elevated-demand regime from a warm-season regime, Gaussian mixture modelling additionally resolved a transitional shoulder-season state, and DBSCAN flagged 17 atypical days, which predominantly coincided with an extreme cold wave, a late-spring frost, low-occupancy cold weekends, and unusually windy days. Electricity-related CO₂ emissions (2,238.7 t), estimated with a fixed grid factor, are reported as a complementary sustainability indicator. The proposed framework converts a single campus-level meter and standard weather data into an interpretable monitoring layer—distinguishing climate-sensitive, operationally driven, and atypical demand days—that is transferable to other district-heated campuses lacking submetering.