The reasonable route choice set in large and complex metro networks; an implementation of the K-shortest path algorithm for the London Underground

Numerous models have been developed to estimate route choice from smart card data. However they require route choice set as an input. In this paper we propose a method to determine that automatically. Firstly, we find the K shortest routes. Secondly, we set cut-off criteria to find the reasonable routes, by comparing the in-vehicle, wait and access egress interchange (AEI) cost of each route with the shortest route’s. We work with 7 origin destination (OD) pairs of the London Underground for which more

routes were surveyed. Among them 5 of them had similar properties and from that it resulted that a route is considered reasonable, if its cost components are up to 1.71, 1.45 and 1.26 times the shortest route’s respectively. From the 2 outliers we understood that we need to include OD specific attributes when setting cut-offs: route with minimum interchange and lines available at origin and destination.