Regret-based models of mobility (451-10-001)
Overview
Background & policy context:
The desire to understand mobility has been a central motivation for the development of so-called discrete choice-theory. Over the years, this Nobel prize-winning econometric modelling approach has shown to be very effective in predicting choices of individual travellers as well as aggregate mobility patterns. The large majority of discrete choice-models are built on utility-maximisation decision rules.
Objectives:
The proposed research pushes the envelope of a new, regret-minimisation based discrete choice-model class recently proposed by the applicant. The approach is based on the premise that people, when choosing, aim to avoid the situation where one or more non-chosen alternatives perform better than the chosen one on one or more attributes.
Compared to utility-based models, the regret-based approach allows for substantial increases in realism and predictive performance, while displaying a high level of econometric tractability. As such, regret-based discrete choice-theory provides a particularly effective modelling approach which may help scholars and policy-makers better understand, predict and manage mobility.
Methodology:
The proposed research aims to provide methodological breakthroughs and substantive insights regarding regret-based modelling of mobility. In particular, it aims at providing the following innovations:
1) Extension of regret-based discrete choice-models towards capturing learning dynamics and choice from ordered alternatives.
2) Derivation of a regret-based measure of accessibility, and a regret-based stochastic user equilibrium in transport networks.
3) Integration of the regret-based approach in transport network-optimisation models.
Derivation of empirical insights into the role of regret minimisation in traveller decision-making and into the potential of the developed regret-based approach as a model of mobility. Data is collected using stated choice-experiments and an interactive travel simulator-experiment.
Share this page