Datasets for occupancy profiles in apartment-style student housing for occupant behavior studies and application in building energy simulation

Abstract

Building energy simulation (BES) tools fail to capture diversity among occupants’ consumption behaviors by using simple and generic occupancy and load profiles, causing uncertainties in simulation predictions. Thus, generating actual occupancy profiles can lead to more accurate and reliable BES predictions. In this article, occupancy profiles for apartment-style student housing are presented from high-resolution monitored occupancy data. A geo-fencing app was designed and installed on the cellphones of 41 volunteer students living in student housing buildings on Clarkson University's campus. Occupants’ entering and exiting activities were recorded every minute from February 4 to May 10, 2018. Recorded events were sorted out for each individual by the date and time of day considering 1 for ‘entered’ events and 0 for ‘exited’ events to show the probability of presence at each time of day. Accounting for excluded days (234 days with errors and uncertainties), 1,096 daily occupancy observations were retained in the dataset. Two methods were used to analyze the dataset and derive weekday and weekend occupancy schedules. A simple averaging method and K-means clustering techniques were performed [1]. This article provides the input datasets that were used for analysis as well as the outputs of both methods. Occupancy schedules are presented separately for each day of a week, weekdays, and weekend days. To show differences in students’ occupancy patterns, occupancy schedules in 7 clusters for weekdays and 3 clusters for weekend days are provided. These datasets can be beneficial for modelers and researchers for either using provided occupancy schedules in BES tools or understanding occupant behaviors in student housing.