Lead Presenter Biography

Bo Wen received his Bachelor and Master of Applied Science in Civil Engineering from the University of Toronto. He is currently a Senior Modeler at TransLink, the statutory authority responsible for the regional transportation planning of Metro Vancouver. He works with the Modeling Forecasting team at TransLink to develop and maintain Metro Vancouver region’s transportation model. His expertise is in the analysis of household travel survey data, travel demand modeling, and geospatial visualization of travel models. As a member of the AI and Advanced Committee at TRB, he is interested in exploring ways to use machine learning and cloud computing to advance the state of practice in transportation modeling.

Co-Authors

Presentation Description

The Regional Transportation Model (RTM) is a four-step travel demand model that currently relies on aggregate segmentation to analyze the demand of different user groups. Developed by TransLink for the Metro Vancouver and Fraser Valley Regional Districts (“our region”), the RTM captures aggregate demand by trip mode, purpose, and household income levels, and has been used in the evaluation of a wide range of transportation/planning projects. With increasing interest in understanding the impact of transportation policies on equity-seeking groups and marginalized communities, our region needs a more advanced modeling system that can track individuals’ travel decisions across many sociodemographic attributes. We considered a variety of modeling paradigms, including tour-based, hybrid trip-based/tour-based, and activity-based models. We determined that activity-based modeling (ABM) was the most suitable approach to address the needs of our region by providing the behavioral resolution needed to understand household- and person-level decisions.
As we embarked on the development of our region’s first activity-based travel demand model, we designed a set of evaluation criteria and weights based on our region’s needs, to aid in the selection of a software platform. These criteria included model performance, user experience, platform maturity, configurability and programmability, development cost, and licensing requirements. We determined that ActivitySim, an open-source platform with development led by a consortium of planning organizations, best met our needs. In the first phase of ABM development, we were able to leverage data processing scripts and example models from peer agencies, such as the Southeast Michigan Council of Governments, Metropolitan Washington Council of Governments, and Chicago Metropolitan Agency for Planning, to develop a synthetic population, to process survey data, and to build an alpha-version model of our region. With the modular design of the ActivitySim platform, we were able to add, remove, reorder, and update sub-models during our development to customize our ABM to better reflect the travel decision processes of our residents. With the alpha-version model as the foundation, we continue to operationalize our ABM in the next phase of development with model estimation, calibration and testing, and supply model integration. 

Presentation File