An ordered probability model for predicting outdoor thermal comfort

Abstract

Outdoor thermal comfort in urban spaces is gaining increasing research attention because it is associated with the quality of life in cities. This paper presents an ordered probability model for predicting the probability distribution of thermal sensation votes (TSVs) based on 1549 observations obtained from a large-scale field survey conducted at a park in Tianjin, China. With a given set of inputs, the developed model can predict the probability that people will feel cold, cool, slightly cool, neutral, slightly warm, warm, or hot. The predictive capability of the ordered probability model was systematically assessed by comparing it with the survey data and a traditional multivariate linear model. Both models had a similar accuracy in predicting single-value TSVs. However, the ordered probability model performed much better than the multivariate linear model in predicting the probability distribution of TSVs. A sensitivity analysis of the ordered probability model revealed that outdoor air temperature was the most important influencing factor. The impacts of global radiation, relative humidity, and activity level on predicted thermal sensation depended on the outdoor air temperature. The developed ordered probability model was used to predict suitable time periods for holding outdoor activities in Tianjin across a whole year. This new model is a more informative tool for predicting outdoor thermal comfort.