Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
Since the 70s, there has been a growing interest in activity-based modelling. This approach models the need to travel as a result of performing daily activities (Bowman and Ben-Akiva, 2001). Nevertheless, the activities need to be scheduled which involves a lot of variables and results in a huge number of unique alternatives (Pougala et al., 2021). Among these variables, the number of possible locations is usually bigger than other variables, motivating the use of a choice set for locations. However, this choice set of locations is usually not known by the modeller (Pagliara and Timmermans, 2009), so there is a need to recreate it. In addition, it would be useful for two purposes: simulation of daily schedules, and estimation of the parameters of an activity-based model based on an underlying choice model. For the first one, alternatives in the choice set must be competitive, to generate realistic schedules, as for the latter one, it should also contain unlikely alternatives to estimate unbiased parameters. In this paper, a methodology to generate a choice set of destinations suitable for both purposes is presented. The choice set is generated with a choice model and can be transformed afterwards to include unlikely alternatives. The methodology is validated using the 2015 Swiss Mobility and Transport Microsensus (fédéral de la statistique and fédéral du développement Territorial, 2017) dataset.
Mario Paolone, Vladimir Sovljanski
Jan Skaloud, Davide Antonio Cucci, Kenneth Joseph Paul
, ,