A Machine Learning Approach for Improved Thermal Comfort Prediction in Sustainable Built Environments

Abstract

Thermal comfort prediction within sustainable built environments stands as a pivotal challenge intertwining human well-being and environmental sustainability. This paper presents a pioneering framework leveraging machine learning methodologies to advance predictive models for thermal comfort. Drawing upon a comprehensive dataset sourced from ASHRAE field studies and the RP-884 database, comprising 107,463 entries, our study unfolds a novel approach to enhancing thermal comfort predictions. The integration of diverse physiological parameters, environmental data, and occupant preferences forms the foundation of our machine learning-driven framework. Through meticulous analysis and model development, our approach not only refines predictive accuracy but also underscores adaptability across varying environmental contexts. The study contributes not only to the discourse on thermal comfort prediction but also emphasizes the crucial nexus between sustainable design, occupant well-being, and energy efficiency.