Hierarchical and K-means clustering to assess thermal dissatisfaction and productivity in university classrooms

Abstract

The thermal environment is one of the main factors that influences Thermal Comfort and, consequently, the productivity of the occupants. The poor quality of the indoor environment in classrooms can be a risk to the health and productivity of students. This study evaluated the relationship between thermal dissatisfaction and productivity in university classrooms through a cluster analysis by hierarchical and non-hierarchical (k-means clustering) methods to verify how thermal dissatisfaction influences students' productivity. Two methods were adopted for determining the actual percentage of dissatisfaction (APD1 and APD2) and productivity (PROD1 and PROD2). After all calculations, 4 clusters were obtained; from the dendrogram, it was observed that clusters 2, 3, and 4 had the solutions with the highest stability. With the formation of clusters and their subsequent characterization, it was possible to observe the relationship between the two productivity methods and other variables, such as thermal dissatisfaction. The most significant variable for separating the clusters was PROD2. A discriminant analysis was performed to confirm that the clusters formed were effective, obtaining a 95.4% success ratio. Thus, when comparing the two most populous clusters, it was found that if self-perceived productivity is the highest, then thermal dissatisfaction based on thermal sensation votes will be the lowest. Also, it is suggested that there was a relation between the time of day and self-perceived productivity. Cold sensation is related to lower self-perceived productivity, while thermal neutrality is related to higher self-perceived productivity.