Human-Building Interactions, Bayesian Estimation Technique
Despite the significant impact of occupant interactions with window shading systems on visual comfort and building energy consumption, there are still significant gaps in understanding and predicting these complex phenomena. This paper presents a Bayesian modeling approach for the prediction of states of motorized roller shades operated by occupants. It is based on a field study with large number of human test-subjects in a high performance building with advanced technology and easy-to-access user interfaces for environmental controls. Unlike the Frequentist methods used in previous studies, the Bayesian approach allows for uncertainty quantification which provides further insight on parameter estimates in models. This information is important when dealing with rather small-sized datasets which is often the case in real applications of human data collection. In addition, this study contributes to the body of knowledge by (1) expanding the investigation of human-building interactions to motorized interior roller shades; (2) considering human attributes and personal characteristics as important underlying variables of window shade use; (3) allowing for prediction of intermediate operating states of the shading system.