Estimation and application of a multi-class multi-criteria mixed paired combinatorial logit model for transport networks analysis

Probabilistic approach of transport network modelling has received significant attention in recent years. Despite the recent progress in this area, the full advantage of the potential capability of random utility choice models has not yet been fully realised. This research is intended to introduce a new approach of combining the state-of-the-art paired combinatorial logit route choice modelling and random coefficient choice models in traffic assignment. While the former addresses the problem of correlation among path utilities, the latter can capture the random taste heterogeneity in route choice decision-making. Including an additional monetary cost explanatory variable, the model would be able to assess a broader range of planning policies such as road pricing. In addition, distinguishing multiple classes of decision-makers, the model has allowed the introduction of demographic aspects of travellers into a network analysis process. Results showed a considerable difference between the flow patterns predicted by the proposed model and the traditional models.