Abstract
Morning peak crowding in urban rail systems is strongly influenced by commuters’ departure time choices, making it essential to understand how commuters respond to congestion and schedule constraints. This study develops a commuter choice model using tap-in and tap-out smartcard data to analyze heterogeneity in departure time preferences and to quantify commuters’ tolerance for deviations from their preferred departure time. The estimation results identify heterogeneous commuter groups characterized as flexible commuters and punctual commuters, reflecting different sensitivities to early and late departures. The analysis further evaluates alternative departure time flexibility intervals that represent the time window within which departures are considered behaviourally equivalent relative to the preferred departure time. Comparisons across interval specifications show that larger intervals improve statistical performance by reducing model complexity, while shorter intervals capture more detailed behavioural variation in commuter departure choices. The results indicate that a 10 minute departure time flexibility interval provides the most appropriate balance between model stability and behavioural interpretability. The estimated alternative-specific constants reveal systematic patterns in departure time preferences, with commuters deriving the highest utility from departures occurring near the early peak period, although many continue to travel during the peak due to external scheduling constraints. These findings provide new behavioural insights into commuter departure time flexibility and highlight the importance of accounting for heterogeneous schedule-delay sensitivities when modelling departure time decisions. The methodological framework offers a robust foundation for analyzing commuter preferences and informing strategies to mitigate peak-period crowding, evaluate peak-spreading policies, and optimize service scheduling in urban rail systems.

This publication has 21 references indexed in Scilit: