Bayesian Inference for Predicting Occupant Satisfaction in Workplace from Indoor Environment Quality

Occupant satisfaction in workplaces is affected by indoor environment quality (IEQ), such as thermal, visual, and indoor air quality (IAQ) parameters. This study aimed to clarify the effects of spatial impressions on satisfaction. We developed a prediction model of satisfaction considering spatial impressions using Bayesian inference. Regression equations for predicting occupant satisfaction from the measured values were derived using a survey dataset from Japan. The occupants were categorized into three groups according to their degree of spatial satisfaction. The posterior distributions of the regression parameters for predicting satisfaction were estimated using PyStan. We adopted the hierarchical Bayesian model, in which parameters are composed of common parts and group differences. The distribution of overall satisfaction predicted from the measured values partially corresponded to that of the true values. It was concluded that occupant satisfaction in workplaces can be predicted by Bayesian inference considering the differences in spatial impression.