Edinburgh Napier University

  With the rapid demand for the energy efficient consumption in buildings, bridging the gap between predicted and measured performance is essential. However, recent studies show that there is a significant mismatch between predicted and actual building performance that is widely known as Performance Gap. In some studies, it is revealed that in-use energy consumption can often be twice as much as anticipated energy consumption. Accurately predicting the energy consumption is a challenging task due to the lack of feedback from occupants’ behavior in post occupancy period. Traditional measurements are not able to simulate and predict the energy consumption precisely and so there is a need for a robust and effective method to overcome such shortcoming. This paper presents a method for predicting the level of comfort in an office building. In this investigation, a boosted regression tree as an artificial intelligence technique from computer science discipline is used to estimate the level of comfort directly from available data in order to achieve a higher accuracy in predictions, a general framework is utilized based on boosting (ensemble of regression trees) that optimizes the sum of square error loss to find the most optimal tree. Furthermore, a Regression Trees (RT) is compared to Boosted Regression Trees (BRT) to show the performance of BRT. According to the experimental results, boosted regression trees provided a powerful analysis tool, giving substantially superior predictive performance to Regression Tree.