Abstract
This study underscores the pressing need to accelerate the energy transition in existing housing to meet the EU’s decarbonization targets. It examines the correlation between thermophysical, technical, and environmental properties of thermal insulation materials and their commercial prices in Greece. Utilizing a dataset of 500 products from technical leaflets, Environmental Product Declarations (EPDs), and market surveys, the study developed six prediction models for material retail prices. A novel aspect of this research lies in quantifying the economic ramifications of insulation material properties through supervised machine learning (ML) techniques. Accurate price prediction models can provide practical applications across stakeholders, including cost estimations for designers, customized recommendations for homeowners, pricing tools for traders, optimization of manufacturing processes, support for policymaking, educational resources for construction professionals, and improved energy efficiency certification payback estimations. Feature analysis showed that total Global Warming Potential (GWP) significantly influences material prices. The Ensemble of Trees and Gaussian Process Regression (GPR) models demonstrated robust performance, with high R2 values of 0.95 in validation and 0.95/0.93 in test sets, indicating reliability without overfitting. Recommendations include integrating more predictors, collaborative model approaches, regular updates, and adopting federated learning techniques for enhanced data privacy and stakeholder cooperation.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call