A dynamic model of time-budget and activity generation: Development and empirical derivation

In this paper we extend and explore a method to estimate dynamic models of activity generation on 1-day travel diary data. Dynamic models predict longitudinal activity patterns of individuals taking into account dynamic needs as well as day-varying preferences and time-budgets. We show how 1-day work time observation probabilities can be derived from proposed dynamic models as a function of a model’s parameters and, with that, how budget-constraints and activity parameters can be estimated using standard loglikelihood estimation.

The results of an application on data from a national travel survey are well interpretable. Moreover longitudinal activity patterns predicted by the model have approximately the same statistical characteristics as the 1-day sample data from the survey. We conclude therefore that the proposed method opens up a way to develop a next generation of dynamic activity-based models of travel demand.