Modelling diversity in building occupant behaviour: a novel statistical approach

We propose an advanced modelling framework to predict the scope and effects of behavioural diversity regarding building occupant actions on window openings, shading devices and lighting. We develop a statistical approach based on generalised linear mixed models to account for the longitudinal nature of observations on occupants, and to provide a coherent method to capture observed variability amongst occupant/room pairings through built-in probabilistic terms describing occupant diversity in a tractable manner within building energy simulation. The contribution of the proposed method is demonstrated using collected behavioural data from three long-term monitoring campaigns (an office building in Switzerland and residential units in Germany and Denmark).