Modelling mobility tool availability at a household and individual level: A case study of Switzerland

Abstract

Prediction of the availability of mobility tools, including vehicles, driving licenses, and public transport subscriptions, is a key task for modelling individual travel and activity-scheduling behaviour. In this paper, we present a new framework for prediction of mobility tool availability, based on sequential logit models of individual driving license ownership, household car ownership, and individual public transport subscriptions. The framework is applied to data derived from the Swiss Mobility and Transport Microcensus (MTMC) to estimate a mobility tool availability model for Switzerland. The utility functions in each sequential model are determined using a machine learning assisted specification approach. The resulting model, which is being incorporated into the Swiss Federal Railways’ nationwide microscopic Agent Based Model (ABM), SIMBA MOBi, is used to simulate mobility tool ownership for a full synthetic population. The predictions are validated against regional and national-level statistics. The results show the new model is able to accurately represent nationally aggregated driving licence ownership by age-group, as well as regionally-aggregated household vehicle ownership and individual half fare travelcard ownership, though further calibration is needed for annual public transport subscriptions. Crucially, the new model will enable the modelling of complex household interactions during activity scheduling and mode choice.