Understanding Travel Behavior Change during COVID-19 Using Spatio-temporal Cluster Analysis

As COVID-19 has been prevalent around the world in recent years, many studies about monitoring and predicting the spread of disease have been conducted in various fields including geography. However, little research has been devoted to infectious disease prediction modeling that adopts constantly changing travel behavior patterns during epidemics. This is due to the limited methodologies to investigate spatio-temporal change in travel behaviors at large-scale and the difficulty in interpreting massive and diverse travel patterns. This study suggests an effective disease surveillance method based on cluster analysis to identify change in travel behaviors during the pandemic by implementing space-time cluster analysis. The results show that K-means++ well represent dynamic changes in travel behaviors at daily scale, whereas retrospective space-time scan statistics have the advantage of detecting travel behavior changes in each period at large spatial scale. Those results could inform decision makers to establish guidelines on travel behavior to curb individual contacts under potential future pandemic.