Abstract
The paper presents an investigation of the temporal transferability of activity generation process models. Three repeated cross-sectional household travel survey data sets collected in the Greater Toronto and Hamilton Area in the years 2001, 2006, and 2011 are used for the investigation. A multiple discrete-continuous extreme value model is used to develop an activity-travel generation model and separate models are estimated for non-workers and workers. Models are developed for individual years and then for the pooled data set of three cross-sectional years to develop a Meta model of activity generation processes. Individual year-specific models are used to increase knowledge about the temporal stability of different parameters of the model so that the Meta model could capture the non-linear evolution of some key parameters of the model. Different transferability indices are used to test temporal transferability of cross-sectional year-specific modes and the Meta model. The results show that, in general, the activity-travel generation process model shows good temporal transferability. The Meta models reveal that the use of multiple repeated cross-sectional data sets considered as a pseudo-panel data improves temporal transferability of an activity generation model significantly.
Acknowledgements
The authors acknowledge the detailed comments and suggestions of four anonymous reviewers.
Funding
The study was funded by an NSERC Discovery Grant and an Early Researcher Award from the Ontario Ministry of Economic Development and Innovation.