“So close, yet so far away”: comparing practices and identifying research opportunities in choice modelling and decision science

Abstract

Choice modellers and decision researchers have largely worked in parallel and with little communication with each other although they share similar goals: to understand and forecast consumer demand, to support decision making and to contribute to the prioritisation between different alternatives. Both fields rely on different types of mathematical models and data sources. Choice modellers use discrete choice models selected for their interpretability while the machine learning methods generally used by decision and business analysts maximise predictive accuracy. Moreover, stated preference methods such as the so-called discrete choice experiment are largely ignored by decision researchers, who usually prefer using multiple criteria decision analysis methods. Why and how should we attempt to bridge this gap is the topic of this talk. Through a series of case studies from different contexts (business, environment, transport, health …), I show concrete examples where better collaboration between fields would be possible and mutually beneficial and I present a research agenda for supporting this objective.

Bio

Romain is a Lecturer in Business Analytics at the Centre for Decision Research, University of Leeds Business School as well as a member of the Choice modelling Centre also at the UoL. He is originally an environmental economists who has also contributed to the transport research literature and the choice modelling literature. His work seeks to improve both stated and revealed preference methods by using new data collection protocols. He has addressed issues such as incentivising truth telling in stated preference surveys and enriching revealed preference data by using new data sources such as smartphone apps.