Model Averaging in Choice Modelling: Applications to model structure and identifying sources of heterogeneity

Stephane Hess, Professor of Choice Modelling, Institute for Transport Studies, & Director of the Choice Modelling Centre, U of Leeds

Tuesday, October 15, 2019, 2:00–3:00pm

Rm. 202, Cummings Hall

Despite the frequent use of model averaging in many disciplines from weather forecasting to health outcomes, it is not yet an idea often considered in travel behaviour or choice modelling. The idea behind model averaging is that a single model can be created by calculating contribution weights for a set of candidate models, depending on their relative performance, thus creating an 'average.' In this presentation, we demonstrate that this idea can be used effectively for travel behaviour modelling.

We identify three key opportunities for model averaging:

  1. For the difficult choice between a number of desirable advanced models;
  2. When advanced models cannot be used due to the size of the data and/or choice sets;  
  3. To help reduce the risk of confounding between heterogeneity in individual sensitivities and heterogeneity in model structures.

Additionally, we demonstrate that model averaging can be used to obtain more reliable elasticities and welfare measures by averaging across outputs obtained from the set of candidate models.

About the Speaker

Stephane Hess

Stephane Hess is Professor of Choice Modelling in the Institute for Transport Studies and Director of the Choice Modelling Centre at the University of Leeds. He is also Honorary Professor in Choice Modelling in the Institute for Transport and Logistics Studies at the University of Sydney, Honorary Professor of Modelling Behaviour in Africa at the University of Cape Town, and has a director position at RSG, a leading North American consultancy company. He has made contributions to the state of the art in the specification, estimation and interpretation of such models, as well as in facilitating the transition of ideas and approaches across disciplines, notably by also working in mathematical psychology and behavioural economics.

Together with his research team at the Choice Modelling Centre, he is setting the research agenda in applying choice modelling in new fields, including education, lifestyle choices, social (network) interactions and joint decision making. Advanced choice models require high quality data, and Hess and his team are leading the field in exploring and exploiting novel data sources, with numerous applications using ‘big data’. Hess has published over 100 peer reviewed journal papers, and his work is highly cited, with a Scopus H-index of 28 (google scholar H-index of 45). He has been involved as a principal or co-investigator in academic projects with a combined income generation of over £9.5M.

His contributions have been recognised by numerous international awards. He is also the founding editor in chief of the Journal of Choice Modelling and the founder and steering committee chair of the International Choice Modelling Conference. He serves on the editorial advisory board of two other leading journals; Transportation Research Part B and Transportation, as well as academic committees at leading international conferences. Together with David Palma, he is the author of Apollo, a highly flexible and powerful free tool for estimating and applying choice models.

For more information, contact Holly Buker at holly.a.buker@dartmouth.edu.