Unsupervised Learning-Based Exploration of Urban Rail Transit Passenger Flow and Travel Pattern Mining

Abstract

This study delves into the realm of urban rail transit systems, leveraging unsupervised learning techniques to analyze passenger flow characteristics and unearth travel patterns. Focused on the dynamic and complex nature of urban rail networks, the research utilizes extensive datasets, primarily derived from Automated Fare Collection (AFC) systems, to provide a comprehensive analysis of passenger behaviors and movement trends. Employing advanced algorithms like DBSCAN, the study categorizes passengers into distinct groups, including tourists, shoppers, thieves, commuters, and station staff. These classifications reveal intricate patterns in travel behaviors, significantly contributing to a deeper understanding of urban transit dynamics. The findings offer valuable insights into peak travel times, popular routes, and station congestion, highlighting potential areas for operational improvements and infrastructure development. The study’s application of unsupervised learning in analyzing vast, unstructured data sets a precedent in urban transportation research, showcasing the potential of artificial intelligence in enhancing the efficiency and sustainability of urban transit systems. The insights garnered are pivotal not only for optimizing current operations but also for shaping future expansion and adaptation strategies, ensuring urban rail systems continue to meet the evolving needs of growing urban populations.