Clustering-based Travel Pattern Recognition in Rail Transportation System Using Automated Fare Collection Data

Passenger travel pattern analysis is essential for the design and development of public transport network. Nowadays, Automated Fare Collection (AFC) systems are widely exploited in the operation and management of public transportation. The data collected from AFC systems provide valuable information to analyze passenger behavior. This research aims to investigate passenger mobility patterns from both temporal and spatial perspectives. We present a hybrid topic-clustering method for extracting travel feature and grouping passengers based on their travel patterns. Our proposed method is illustrated using a real AFC dataset of the metro transportation system in Shenzhen, China. The results showed that four temporal travel patterns were well identified. Comparison of travel behavior indicated that metro travelers with different travel time selections also have different activity areas.