Abstract
Across the globe, buildings account for a substantial portion of energy consumption and CO2 emissions. In this paper, we use time series-based machine learning models to predict building occupancy and control building heating, ventilation, and air conditioning (HVAC) for energy savings. We achieve a 7.4% reduction in daily HVAC energy usage in our target building. Our approach leverages existing wireless access point (WAP) infrastructure to gather ground-truth occupancy data in a privacy-conscious way. Our method can be used across multiple buildings and can be expanded to different use cases.