Research Output

Estimation of weights of times and transfers for hyperpath travellers.

  In public transport networks with high frequency services the travellers are often assumed to reduce their travel time by identifying for each node a set of attractive lines, each of which might be the fastest from the node, depending on its and other lines arrival time. Almost all transit assignment models have been developed using this concept. The question of valuing transfer penalties and waiting times compared to on-board travel times has been largely ignored. Contrary, the literature estimating values of times for public transport passengers largely ignores the complexity of choices faced by travellers, especially in networks where passengers can choose several lines from a stop. This paper therefore attempts to close this gap by addressing questions such as: Do different passenger groups choose different strategies at stops? Given this objective a web-based survey was conducted. By using these data, hyperpath selection is formulated as a discrete choice model and the relative weights are estimated. A challenge for this study is the consideration of the influence of correlations among alternatives. A hyperpath is by definition a set of elementary paths which minimises the expected travel time, and it is likely that the same lines are part of different hyperpaths. Therefore the assumption of independent and irrelevant attributes does not hold. To overcome this problem, we attempt to estimate the parameters by nested logit models. Through the comparison of the estimation results, the advantage of considering correlation is discussed.

  • Type:

    Conference Paper (unpublished)

  • Publication Status:


  • Library of Congress:

    HE Transportation and Communications

  • Dewey Decimal Classification:

    388 Transportation; ground transportation


Schmoecker, J., Kurauchi, F., Fonzone, A., Hemdan, S. M. H., Shimamoto, H. & Bell, M. G. H. Estimation of weights of times and transfers for hyperpath travellers.. Paper presented at 91st Annual TRB Meeting, Washington, D.C., USA. (Unpublished)



Hyperpath choice; weight of times; stated preference study; cross nested logit model;

Available Documents