Occupants influence thermal behavior of buildings due to their presence (e.g., via releasing sensible and latent heat), and via operation of control devices such as windows, shades, luminaries, radiators and fans (Mahdavi 2011). Specifically, knowledge of occupants’presence represents a necessary condition for the development of predictive control action models. Performance simulation tool users typically deploy libraries of diversity factors and schedules to represent occupants’ presence in buildings. These diversity profiles are derived from long term monitored data in different classes of buildings and are usually included in the simulation packages to facilitate the creation of building performance models. More recently, efforts are being made in the scientific and professional communities to develop probabilistic models that would capture the randomness of occupants’ presence. As one of the first attempts, Newsham et al. (1995) considered the probabilistic nature of occupancy while developing a stochastic model to predict lighting profiles for a typical office. Their model deployed the probability of first arrival and last departure as well as the probability of intermediate leaving and returning. Reinhart (2001) further developed this model by using the inverse transform method (Zio 2013) to generate

samples from the distribution functions of arrival and departure times. Moreover, days were divided into three phases (morning, lunch and afternoon) for which the probabilities of start time and length of breaks were computed. Page et al. (2008) proposed a generalized stochastic model for the simulation of occupants’ presence using the presence probability over a typical week and a parameter of mobility (defined as the ratio of state change probability to state persistence probability). They also included long absence periods (corresponding to business trips, leaves due to sickness, holidays, etc.) as another random component in their model.