TRB 2016 Blue Ribbon Committee
16th National Transportation Planning Applications Conference

Validating Trip Distribution in Southeast Michigan using GPS Data

Corresponding Author: Sean McAtee, Cambridge Systematics, Inc.

Presented By: Sean McAtee, Cambridge Systematics, Inc.


It is critically important that a travel demand model accurately predicts how trips are distributed throughout a region. Traditionally, datasets such as household travel surveys, Census Transportation Planning Products (CTPP, and Longitudinal Employer-Household Dynamics (LEHD) data have been used to support calibration of trip distribution models. More recently, “Big Data” obtained from cell phone and GPS devices has been used to supplement information about trip distribution patterns. This presentation will explore use of StreetLight data to improve the regional trip distribution model for the Southeast Michigan Council of Governments (SEMCOG). This data source provides information for both personal and commercial vehicles, allowing analysis to be conducted separately for each type of travel.

For personal travel, StreetLight data was compared to the CTPP worker flow data and two household travel surveys conducted in 2005 and 2015. The household travel surveys have been weighted to account for variability in sampling rates for different demographic groups, while the StreetLight data has not been weighted. Evaluation of the un-weighted data suggested that demographic characteristics affect sampling rates in the StreetLight data. The presentation will show evaluation of various expansion approaches for their ability to reduce sampling bias while preserving rather than obfuscating trip patterns present in the original data. Opportunities and limitations encountered using StreetLight as a supplement to traditional data sources will then be identified. In particular, the StreetLight data provides information about external trips that are difficult to observe, as well as information on difficult to reach traveler segments including visitors to the region by providing data on trips starting at the airport and at external stations.

StreetLight data also serves as a source of commercial trip data. The StreetLight commercial vehicle data was more difficult to evaluate than personal travel data due to a lack of other sources of observed data in the SEMCOG region. For this analysis, commercial vehicle trip patterns in the StreetLight data were compared to trip patterns obtained through survey efforts at the Canada-USA border conducted by Transport Canada. This comparison leads to adjustments to data processing and analysis methodology.


Discuss This Abstract