Global regionalization of heat environment quality perception based on K-means clustering and Google trends data

Abstract

To effectively plan for the thermal environment in the face of climate change, it is crucial to consider regionalized approaches and people's perceptions of the phenomenon based on actual experiences. This study performs perception-based regionalization research of the thermal environment using Google Trends search query volume data. Global Google Trends data for 12 terms related to the thermal environment were collected from 2016 to 2022 and analyzed by time series and geographical units. The study found that the correlation between geographical unit data was higher than that of the time series units. To propose a global regionalization map, we used K-means clustering on the geographical Google Trends dataset and determined the optimal number of five clusters using the elbow method. Through a detailed analysis of each term for derived clusters A to E, the study revealed findings and implications that would contribute to the literature on the thermal environment. Finally, the perception-based global regionalization map was proposed. Overall, this novel approach to determining global regions based on people's perceptions of the thermal environment with Google Trends data provides insights for effective future thermal environment planning by analyzing the priority of characteristic groups and indicators by relevant regions for each cluster.