Non-invasive Procedure to Probe the Route Choices of Commuters in Rail Transit Systems

Accurately determining the probability of various route choices is critical in understanding the actual spatiotemporal flow of commuters and the instantaneous capacity of trains and stations. Here, we report a novel procedure, based solely on the recorded tap-in tap-out ticketing data, that dictates the route choice of commuters in a rail transit system (RTS). We show that there exists a signature travel time distribution, in the form of Gumbel type 1 function, from a given origin O to a destination D. Any particular route can then be considered as a superposition of this mapping function and one can compute the probability that a specific path, over other possible paths, is taken by a commuter from O to D. The procedure is demonstrated by considering different scenarios using travel data from smart fare cards from Singapore's RTS; results show that the forecasted characteristic profile deviates by less than 10−5 from the actual distribution. We note that our method utilizes only two parameters that can be experimentally accounted for