Machine learning based prediction of thermal comfort in buildings of equatorial Singapore

Majority of energy consumption in Singapore buildings is due to air-conditioning, because of its hot and humid weather. Besides attaining a healthy indoor environment, a prior knowledge about the occupant's thermal comfort can be beneficial in reducing energy consumption, as it can save energy which is otherwise spent in extra cooling. This paper proposes a data-driven approach to predict individual thermal comfort level (`cool-discomfort', `comfort', `warm-discomfort') using environmental and human factors as input. Six types of classifiers have been implemented-Support Vector Machine (SVM), Artificial Neural Network (ANN), Logistic Regression (LR), Linear Discriminant Analysis (LDA), K-Nearest Neighbors (KNN), and Classification Trees (CT), on a publicly available database of 817 occupants for air-conditioned and free-running buildings separately. Results show that our approach achieves prediction accuracies of 73.14-81.2%, outperforming the traditional Fanger's PMV (Predicted Mean Vote) model, which has accuracies of only 41.68-65.5%. Age, gender, and outdoor effective temperature, which are not included in the PMV model, are found to be important factors for thermal comfort. The proposed approach also outperforms modified PMV models-the extended PMV model and the adaptive PMV model which attain accuracies of 61.75% and 35.51% respectively.