Capturing correlation in large-scale route choice models

When using random utility models for a route choice problem, choice set generation and correlation among alternatives are two issues that make the modeling complex. In this paper we discuss different models capturing path overlap. First, we analyze several formulations of the Path Size Logit model proposed in the literature and show that the original formulation should be used. Second, we propose a modeling approach where the path overlap is captured with a subnetwork. A subnetwork is a simplification of the road network only containing easy identifiable and behaviorally relevant roads. In practice, the subnetwork can easily be defined based on the route network hierarchy. We propose a model where the subnetwork is used for defining the correlation structure of the choice model. The motivation is to explicitly capture the most important correlation without considerably increasing the model complexity. We present estimation results of a factor analytic specification of a mixture of Multinomial Logit model, where the correlation among paths is captured both by a Path Size attribute and error components. The estimation is based on a GPS dataset collected in the Swedish city of Borlänge. The results show a significant increase in model fit for the Error Component model compared to a Path Size Logit model. Moreover, the correlation parameters are significant