Exploring the spatiotemporal characteristics of intracity travels by time series clustering: A case study of Beijing

Abstract

AbstractExploring spatiotemporal characteristics of residents' travel is a fundamental task in the field of urban studies. Time series clustering is one of the useful methods to distinguish those characteristics of intracity travel. However, previous studies focused more on exploring the spatiotemporal characteristics in urban centers. For understanding the spatiotemporal characteristics in suburbs, we combined the Shape Based Distance (SBD) based on the Discrete Fourier Transformation (DFT) similarity measurement method and the Affinity Propagation (AP) clustering method to cluster time series of each Traffic Analyze Zone (TAZ). The method performed well both in urban centers and suburbs. There are nine clusters on weekdays and seven clusters on weekends. These clusters reflect different travel characteristics. To further understand the difference between these travel characteristics, Points of Interest (POI) data were employed to analyze attractors of travel. Work‐oriented, Residence‐oriented and Attraction‐oriented are three main attractors on weekdays. Meanwhile, there are four attractors on weekends which are Recreation‐oriented, Residence‐oriented, Attraction‐oriented and Work‐oriented. This research has a certain theoretical and practical value for understanding urban formation evolution and urban planning.