Computational predictions of buildings' indoor-environmental conditions and energy performance would presumably benefit from the inclusion of models that could reliably capture occupants' window operation behaviour. Frequently, models derived from empirical data have a black-box character. However, the utility of window operation models could be conceivably improved, if the model derivation process is preceded by specific hypotheses regarding the variables that are assumed to influence the frequency and timing of window operation actions. In the present contribution, we discuss the process of exploring explicit hypotheses regarding factors that could influence occupants' operation of windows prior to the model derivation step. To illustrate the potential of this approach, we utilize a specific window operation data set from an open plan office. This data set was used to test three distinct hypotheses regarding the factors that influence occupants' window operation actions upon arrival. The results suggest that the most plausible conjecture from the intuitive point of view is not supported by the data set. This observation encourages more in-depth reflections on the motivational background of occupants' behaviour. Purely data-driven black-box models arguably do not provide a similarly strong impetus toward an explicit understanding of occupants' behaviour patterns in buildings.
