Modelling the impact of activity duration on utility-based scheduling decisions: a comparative analysis

Abstract

There exist two major categories of activity-based models: on one side, utility-based or econometric models are founded on the principles of random utility maximisation, and use discrete choice modelling techniques to solve activity-based problems. On the other hand, rule based approaches refute the assumption that decision-makers are perfect optimisers and their activity-travel behaviour is the product of context-dependent rules. Recently hybrid models (e.g. OASIS) combine both approaches, to keep the flexibility and theoretical robustness of utility-based models, with the addition of spatio-temporal constraints which increase the behavioural realism and simplify the estimations. However, hybrid models suffer from two main issues: specifying a utility function that accurately reflects the decision-making process, and estimating parameters in a highly complex space due to the constraints. In this paper, we answer these questions within the context of the OASIS framework (Pougala et al., 2022). We first estimate the parameters of two state-ofthe-art utility functions (Charypar & Nagel, 2005; Feil, 2010), using data from the Swiss Mobility and Transport Microcensus (Office fédéral de la statistique and Office fédéral du développement Territorial, 2017) and compare them with OASIS’ default linear-in-parameters utility function of the model.