Benefits of Automating Calibration/Validation of Oregon MPOs Travel Demand Models
Corresponding Author: Jinxiang Ren, Oregon Department of Transportation
Presented By: Jinxiang Ren, Oregon Department of Transportation
Abstract
Oregon Department of Transportation (ODOT) develops, maintains, and applies travel demand models for 5 small- and medium-sized metropolitan planning organizations (MPOs) and 15 small urban areas (cities and counties). The large number of models requires ODOT Transportation Planning and Analysis Unit (TPAU), the agency to be resource efficient. Faced with these challenges, TPAU developed two sets of transferable modeling structures: JEMnR (Joint Estimation Model in R) for MPOs and OSUM (Oregon Small Urban Models) for non-MPOs. Even though this approach removes the burden of constructing different model structures for different urban areas, still the standard calibration/validation process needs to be undertaken. In 2010-2011, Oregon conducted statewide household activity surveys that would enable us to revisit, reassess and/or readjust the model coefficients and rates.
To address the MPOs’ modeling needs, TPAU opted for a practical way by implementing effective and flexible scripting approach to automate the step-wise calibration/validation processes for travel demand models, such as: household sub-models, trip generation, destination choices, and mode choices. The scripting allowed us to easily revisit these calibration steps identified in a peer review process in a quick way. Scripting’s main benefits are summarized in the following:
? Repeatability of processes
? Transferability of coding practices, checks and balances
? Rapid trial and error runs
? Efficiency in numerous iterations
? Exploration of data and methodologies
? Ability to deal with full and thin observed data
? Flexible and responsive tool implementation in all three MPOs
? Variations for every model component
? … and it’s fun!