Pougala, J., Hillel, T., and Bierlaire, M. (2020)

An optimization framework for daily activity schedules

9th Symposium of the European Association for Research in Transportation, Lyon, France

Transport planners and operators have to face nowadays increasingly complex mobility behaviors. Traditional trip-based models become very limited in terms of behavioral accuracy when it comes to anticipating and accommodating these new, and hard-to-capture needs. The shift towards activity-based approaches is thus natural, as this alternative is better equipped to deal with individual-level granularity. The assumption behind these models is that all transport-related choices made by a person (e.g. number of trips, location and mode choice) are derived from the need to do activities, and their spatio-temporal sequence. We propose a modelling approach based on first principles: a traveler schedules their activities in order to maximize the total utility they can derive out of them, thus solving a mixed integer optimization problem. Our model allows to generate distributions of schedules for each individual, from which we can draw likely outcomes. Another contribution is the simultaneous inclusion of multiple choice dimensions (e.g. activity, location, mode choices. . . ), allowing for more flexibility than current models that treat them sequentially. The model was tested using trip diary data from the Swiss Mobility and Transport Microcensus. The results show that we are able to generate realistic activity schedules for a wide range of individuals