Development of artificial neural network models for predicting thermal comfort evaluation in urban parks in summer and winter

Abstract This study has successfully formulated artificial neural network models to predict thermal comfort evaluation in outdoor urban parks in Hong Kong, a sub-tropical city, for both summer and winter periods. The artificial neural network models embracing two-hidden layers outperformed other types of commonly adopted thermal comfort models. The model prediction performance was considerably improved by including perceptions of microclimate, perceptions of environmental features and personal traits as additional predictor variables. Sensitivity analysis determined that thermal sensation is the most important factor influencing thermal comfort evaluation in outdoor urban parks, followed by air temperature for both summer and winter. Solar radiation is another important factor immediately following air temperature for winter. In contrast, perceived density of trees and perceived number of water bodies in a park were found to be more important than solar radiation for summer. The findings arising from this study should provide valuable insights for formulating effective strategies for improving the thermal environment in urban parks in different seasons.