Dataset for Context-Based Modeling and Prediction of Energy Consumption in Sofia City

This dataset supports the publication Dataset for Context-Based Modelling and Prediction of Energy Consumption in Sofia City. It contains a cleaned, geolocated point dataset of Energy Performance Certificate (EPC) records for buildings in Sofia, Bulgaria. The dataset includes validated EPC attributes, along with geographic coordinates derived from cadastral ID matching, address normalisation, and UPI-based geolocation.All records were systematically cleaned to remove duplicates, inconsistent entries, missing core attributes and incorrectly located points. The dataset enables spatial analysis of building-level energy performance, integration with additional urban datasets, and development of machine learning models for energy consumption prediction.The footprint-linked dataset used internally for modelling is not included due to structural limitations in the cadastral data (for example, multifunctional buildings represented as single polygons or newly constructed buildings missing from the cadastre). Users may reproduce the footprint-linking step using this point layer and the official cadastral data for Sofia Municipality.This dataset can be used for:• Spatial analysis of building energy performance in Sofia• Urban energy modelling and planning• Machine learning–based prediction of energy demand• Integration with morphology, land-use and urban analytics datasets File format: GeoPackage (.gpkg)Total records: 2027 Developed within the internal research programme on urban energy modelling at the GATE Institute.