RUS boost tree ensemble classifiers for occupancy detection

Abstract

In this research paper, various ensemble classifiers are used to predict occupancy status using samples of light, temperature, humidity, CO2, humidity ratio sensor data. Occupancy detection will save energy making room for smart buildings in smart cities. It paves ways to decide on heating, ventilation, cooling and lighting. To achieve 'white box' output and facilitate explanatory interpretation, decision tree was employed, Several weak learner decision trees were melded to form RUSBoosted Tree ensemble classifier. On investigation of the results, it is seen that RUSBoostedTree Ensemble gives the highest accuracy rate of 99%.