Identification of Window Openness in Smart Buildings by Random Forest Algorithm

Controlled energy distribution and energy efficiency in buildings are among the main concerns in modern constructions, energy management and buildings usability. Total energy performance of a building depends on several factors, which include materials and components, building environment, but also the occupants behavior. User interaction with windows and window status as open or closed is the main cause of the difference between the predicted and the actual energy consumption. Understanding these user habits helps to better predict energy disruptions and enables planning for more efficient energy distribution. This paper investigates the possibility of using a machine learning model to identify the user behavior through a set of available historical data in a living-lab smart building. The random forest algorithm used in this research proved promising for the particular use case and makes a good ground for future work with window status classification accuracy of 87.85%.