Alternative Data Sources for Transportation Planning

Click to enlarge

Sponsors:

National Science Foundation Graduate Research Fellowship
Federal Highway Administration Eisenhower Graduate Research Fellowship

Students:

Josie Kressner
Greg Macfarlane
Stefan Binder

In 2011, the Atlanta Regional Commission (ARC) spent more than $2 million to survey the travel patterns of approximately 10,000 households in its 20-county planning area [1]; this survey is the primary source of data for the region’s travel demand models. Travel models are a system of smaller models that predict auto ownership, trip making, route choice, and other variables. This modeling system helps quantify impacts associated with proposed transportation investments or policies and helps metropolitan planning organizations (MPOs) achieve goals in regional mobility and air quality attainment. Because so many decisions are based on the results of travel models, federal agencies may withhold funding from cities that do not regularly update their travel models.

As a group, the 350 MPOs in the United States annually spend tens of millions of dollars to collect travel survey data. Due to these costs, there has been increasing interest in investigating alternative  sources of data for travel demand modeling. For example, the Transportation Research Board of the National Academies has recognized the need for “affordable and sustainable data collection systems for personal travel data” and has noted that “new approaches for collecting data need to be assessed, [with] costs evaluated, and benefits documented.” [2]

As part of their resesarch, doctoral students Josie Kressner and Greg Macfarlane and master's student Stefan Binder investigated the viability of using alternative data sources, such as credit reporting databases, and combining it with other passively-collected third party datasets.  Credit reporting data contain a rich set of individual-level demographic and socioeconomic information.  Unlike census data or household travel surveys, credit reporting data are updated frequently, often on a monthly or quarterly basis, and contains information on lifestyle segmentation variables. Lifestyle segmentation variables consides behavioral and attitudional preferences in addition to demographic and economic characteristics. As people age and household structures change over time, behavioral preferences change. The work by Binder, Kressner and Macfarlane provides insights into the ability to use credit reporting data and other third party data to model vehicle emission failures, automobile ownership decisions, and residential location choices.  Their research provides the foundation for regional councils potentially saving millions of dollars by eliminating or reducing surveys that duplicate relatively cheap information available from credit reporting agencies.

References

[1] Guy Rousseau. Atlanta Household Travel Survey. Presented to the ARC TCC Meeting, October 8, 2010.
[2] TRB. Social Research Needs. Transportation Research Board, 025(2):1–16, 2010.

Publications resulting from this research:

Binder, S., Macfarlane, G., Garrow, L.A. and Bierlaire, M. (2014). Associations among household characteristics, vehicle characteristics and emission failures: An application of targeted marketing data. Transportation Research Part A 59: 122-133. 

Kressner, J. and Garrow, L. (2014). Using big data for travel demand modeling: A comparison of targeted marketing, Census, and household travel survey data for Atlanta. Transportation Research Record 2442(1): 8-19.

Macfarlane, G., Garrow, L.A. and Mokhtarian, P.  (2015). The influences of past and present residential locations on vehicle ownership decisions.  Transportation Research Part A 74: 186-200.