Abstract
The data-driven approach to building evaluation is gaining significant popularity. Policies related to the adoption of new energy-saving technologies are now rooted in their actual effects on reducing energy consumption. The pandemic and subsequent war have emphasized the crucial significance of various building aspects, including air quality, indoor comfort (thermal, light, and acoustic), and energy efficiency. Evaluating these aspects necessitates the use of real data, which, unfortunately, is often inaccessible, of low quality, or incomplete. Yet, even in the energy and building fields, data anonymization methods like normalization and aggregation limit the information that can be shared. Therefore, finding more effective ways to collaborate on data without compromising privacy is critical for both data owners and data analysts. According to the European Commission’s Joint Research Centre, synthetic data will play a crucial role in enabling AI. This type of data serves as a unifying bridge between policy support and computational models, unlocking the potential of data that may be hidden in silos. Consequently, it becomes the primary catalyst for AI adoption in business and policy applications across Europe. Additionally, the resulting data not only can be freely shared but also aids in rebalancing under-represented classes in research studies through over-sampling, making it an ideal input for machine learning and AI models. One of the fundamental elements involves developing AI models that can accurately and faithfully reproduce various types of building data, considering factors such as climatic conditions, building and system types, occupancy profiles, and intended use. The present study aims to evaluate potential scenarios and applications of AI models for generating valuable data for both building energy assessments and economic evaluations.