Development of Smart Indoor Workplace System Using Decision Tree Algorithm

Commonly controlled factors in smart buildings are heating, ventilation, air conditioning and lighting mainly due to optimization requirements. Controlling the physical environment is a dominant factor not only in smart buildings, but also in worker’s productivity. This work aims to develop a system that correlates the occupant work efficiency and the physical environment using machine learning. The system consists of a hardware system for data gathering and a data processing unit which help create and employ a prediction model that correlates the work efficiency of an occupant and the air conditioning and luminance levels in an indoor workplace. The input of the system monitors the ambient lighting, temperature conditions, along with the sitting behavior of the occupants in the workplace and typing job. From the gathered dataset, a machine learning model is produced using decision tree algorithm. All sensor data and predicted outputs are sent to a cloud server and can be accessed remotely through a web interface. Results shows that the prediction model achieved an area under the receiver operating characteristic curve of 0.89 for air conditioning setting and 0.50 for the light setting which shows good and random prediction performance, respectively.