Leila Dianat, PhD thesis, Civil Engineering, 2018
Professor Eric J. Miller and Professor Khandker Nurul Habib, Co-supervisors
This thesis reports on significant progress in several aspects of activity-based travel demand modelling.
The major contribution is extending the time-frame of the activity-based models by developing two week-long travel/activity patterns prediction models. The two models differ in the way they schedule work activities:
- In the first model, work, school and night sleep are considered as the skeleton schedule around which non-work/school activities are organized. Weekly work activities are scheduled in a separate framework applying the concept of a “project” which consists of work episodes which are planned prior to the week (preplanned) and episodes planned during the week (unplanned). These episodes are generated and scheduled in a two-level, weekly, dynamic framework.
- In the second model, night sleep is the only skeleton schedule and the rest of the activities including work and school are generated and scheduled simultaneously. The results indicate the significance of developing models with longer planning horizon than a typical weekday.
The other main contribution of the thesis is investigating the influence of a fundamental assumption in activity-based modelling which is assigning a higher priority to scheduling work/school activities on the predicted travel/activity pattern of the nonwork/school activities.
The thesis also tests the forecasting ability of the developed models, something which is rarely done in the literature.
Supervisor contact information
Professor Eric J. Miller
Tel: (416) 978-4076
Professor Khandker Nurul Habib
Tel: (416) 946-8027