Exploration of the relationships between perceived and observed parameters of IEQ using Bayesian analysis

Abstract

Seeking to bridge the gap between observations and predictions of thermal comfort, recent work has explored novel predictive frameworks to improve the prediction accuracy of occupants’ thermal satisfaction in office spaces. Recent contributions include the development of a Bayesian framework to estimate the probability of an occupant feeling thermally satisfied as a function of not only psychrometric IEQ parameters but also non-thermal metrics of IEQ. A predictive relationship between indoor CO2 concentrations and thermal satisfaction was found, though the underlying causal relationship is not yet clear. An occupant unhappy about air quality is more likely than not found to be unhappy with other parameters, including indoor air temperature. To quantify these relationships, further analysis with new modelling methods and data is required. This paper presents a new formulation of prior work, using a new Bayesian logistic regression model and counterfactual inference to assess the combined relationships between many subjective and objective IEQ factors. This work sets out to provide the first-known Bayesian analysis of the underlying causality of observed statistical relationships between divergent parameters of subjective and objective IEQ.