Predictability quantification of occupant presence in high-rise residential apartment buildings

Abstract

In recent decades, many studies have attempted to develop a reliable occupancy model using either rule-based, stochastic, data-driven, or agent-based approaches. These are based on the hypothesis that occupant presence can become predictable provided sufficient knowledge and data are provided. However, a different view propounds that occupant presence could follow a random-walk pattern or become unpredictable in certain types of rooms/buildings, for example, university labs and library buildings. In this study, the authors report the predictability of occupant presence in high-rise residential apartment buildings in South Korea. The authors collected occupant presence data from 31 households over 147 days using occupancy sensors installed in each household. The predictability of occupant presence was then analyzed using the normalized cumulative periodogram (NCP) and Bartlett’s test. It was found that (1) the predictability of occupant presence is significantly influenced by temporal and spatial resolutions, (2) extending measurement periods (e.g., 7 days vs 147 days) can increase the predictability of occupant presence, (3) for a measurement period of 7 days, the occupant presence for 14 households became unpredictable, and (4) the predictability of occupant presence significantly differs among 31 households.