Clustering analysis of vehicle traveling and parking behavior based on license plate recognition data: Suzhou case

The regularity of vehicle travel patterns is a crucial characteristic of urban traffic operations. Analyzing these patterns yields valuable insights for traffic management and urban planning. With the advent of big data technology and intelligent transportation systems, accurately identifying traffic demands and profiling traveler behavior has become practical. In this study, we leveraged license plate recognition data to cluster and analyze the travel characteristics of vehicles in Suzhou city. Initially, the Suzhou license plate recognition data underwent preprocessing to rectify errors and eliminate redundancies, thereby enhancing data accuracy. Subsequently, a threshold segmentation technique was applied to partition the travel sequences. Finally, a suite of indicators, including parking habits, driving behavior, and driver attributes, were extracted. Travelers were then categorized using the K-means++ clustering algorithm to discern the characteristics of different vehicle user groups. The findings suggest that: (1) Low-frequency foreign vehicles constitute 82.36% of the total foreign vehicle population and contribute 46.11% to the overall foreign travel intensity. High-frequency foreign vehicles, which make up only 2.72% of the total foreign vehicle count, significantly contribute 24.6% to the foreign vehicle travel intensity, highlighting their importance as users of road resources. (2) Travelers are classified into four categories: Light Commuters, Daily Commuters, Weekend Explorers, and Balanced Travelers. Notably, Light Commuters and Daily Commuters display marked differences in both their travel and parking behaviors. This research offers policy recommendations tailored to different vehicle user groups and provides data-driven insights for addressing urban transportation challenges.