Girei,Z.J.B.Chukwumauchegbu,M.I.Adewolu,A.O.Naibi,A.U.Uwa,J.N.2024-09-112024-09-11202401660-679510.62441/nano-ntp.v20iS1.472-s2.0-85195379160https://doi.org/10.62441/nano-ntp.v20iS1.47https://hdl.handle.net/20.500.14517/6311Non-intrusive indoor environment sensing for occupancy detection and estimation has attracted extensive research interest in the building domain over the past decade due to the increasing number of applications for improving building infrastructure. Occupancy detection and estimation can be integrated into building appliances to manage lighting applications, intrusion detection in secured building areas, and occupancy-driven ventilation which has the potential to improve the performance of the Heating Ventilation and air-conditioning (HVAC) system through the finegrained occupant count to enhance the trade-off between thermal comfort and energy consumption. The research strategies for occupancy detection and estimation have utilized different technologies (including camera, wearable, and indoor environmental variables sensing through direct sensing and machine learning), which experience challenges in terms of acquiring essential sensory data related to occupancy information and correctly modeling the occupancy data due to hardware deployment limitations and underlying cost. This study explores existing technologies and methods for occupancy detection and estimation with their underlying issues. It provides a comprehensive procedure for occupancy modeling methodology using different machine learning methods and analyzing their comparative results to assist in decision making for choosing an optimal technique for solving occupancy detection and estimation problem. The results recommend Random Forest as a candidate model with high performance achieving 73.6% to 99.7% for occupancy detection and overall, 99.3% for occupancy estimation. © 2024, Collegium Basilea. All rights reserved.eninfo:eu-repo/semantics/openAccessBuilding energycomponentMachine learningOccupancyRandom ForestArticleQ420S1630639