Investigating consistency in transit passenger arrivals: insights from longitudinal AFC data

Previous research has established that one category of transit passengers 'target' their arrival at a time shortly before the service is scheduled. Another group of passengers arrive randomly, seemingly unaffected by timetable schedules. The time of day plus transit service characteristics, such as headway and reliability, affect the split of passengers between these groups. Little is known about the longitudinal aspects of the non-random arrival behavior. This paper assesses the day-to-day consistency of transit users' arrival behavior by analysing longitudinal automated fare collection system (AFC) data for heavy rail services in Melbourne, Australia. It makes further distinctions of the transit users that have up until now been categorized into random or non-random arrival behavior. Four archetypal arrival behaviors are derived. Consistency of arrival behavior is quantified and investigated. Heterogeneity is an overriding feature of transit commuters' longitudinal behavior. Transit users exhibiting a greater amount of consistency were found to arrive closer to the transit's timetabled time, used fewer scheduled services and traveled earlier in the peak. A systematic difference in arrival behavior was found for users of a terminus station.