A systematic review and comprehensive analysis of building occupancy prediction

Buildings account for a significant portion of the global energy consumption. Forecasting personnel occupancy is critical for reducing energy consumption in buildings. This study explored the general process of building occupancy prediction models, and specifically analyzed the evolution and application of various data collection methods and predictive algorithms. A comprehensive research framework is established. The main findings indicate that prediction accuracy can be substantially improved by leveraging the Internet of Things technology to enhance data collection and employing hybrid machine learning algorithms. These advancements are vital to optimize building operation strategies, reduce energy consumption, and minimize carbon dioxide emissions. Additionally, the assessment metrics for validating predictive models are discussed and a novel idea based on integrated selection methods is presented. Differing from existing research, this study explores data collection methods and predictive algorithms from a broader perspective, also examining their interplay. Finally, potential directions for further development and improvement in the field are identified. The findings emphasize the necessity to continually innovate in data collection and algorithm development to meet evolving environmental needs and sustainability goals. New insights for engineering design and energy system optimization are offered.