Cell Phone Data and Travel Behavior Research: Symposium Summary Report

The Exploratory Advanced Research Program

Linking Trip Information to Behavior Data

In this session, researchers from various agencies and academic institutions provided descriptions of the work that they have been conducting and their uses of linking cellular and other mobile data with behavior and environment data. Symposium participants then discussed challenges relating to land use and margins of error.

Presentations

Lei Zhang, University of Maryland
Lei Zhang’s focus has been to research trip purpose by using passively collected data and by attaining varying trip characteristics and details depending on the model. Zhang linked the 1995 American Travel Survey and cell data for urban trips. Zhang found that the most difficult types of trips to categorize were personal in nature, because there is less of a pattern. His model utilized land use, time of day, and duration and works fairly well across regions. Zhang noted that a useful next step would be linking the data in terms of a time series.

Billy Bachman, Westat
Billy Bachman focused on what can be learned from GPS data in conjunction with travel surveys. He highlighted that developing the link to behavior is important for building forecasts and determining future movement. A challenge that Bachman noted included ensuring that the data stream is uninterrupted or does not contain missing data. This is a specific area in which cellular data could prove to be beneficial. Integrating cellular data into traditional survey approaches could be particularly beneficial in filling in gaps and answering lingering questions.

Shan Jiang, Massachusetts Institute of Technology
Shan Jiang worked on extracting data from cellular devices to link trips to activity. This included looking at possible nodes for daily travel, which typically consisted of home and work. In addition, Jiang linked the data to census tract data with the goal of determining the most heavily impacted roadways in terms of traffic to target congestion reduction, as shown in figure 6. Jiang noted that a key challenge is validating cell phone inferred results with survey and GPS data.

Figure 6. Road usage based on cell phone data linked with census tracts to illustrate congestion exposure.

Major Themes Discussed

The symposium participants discussed several major themes during this session, as follows:

• Improving accuracy. Cellular data are generally less accurate than GPS-tracking data in terms of position. Cellular data focuses mainly on travel among larger zones and monitors the handoff among different cell coverage areas, whereas GPS data can pinpoint specific OD locations.
• Validating results. Multiple data along the same paths assist in validating results by using cellular data.
• Processing issues. Pre-processing of cell data by the telecommunication companies that collect the data and post-processing by data providers can leave artifacts that limit use for understanding travel behavior.
• Developing a nationwide standard. Land-use models and data are extremely helpful, but specificity and configuration varies from region to region, and there is no nationwide standard land-use model.

Key Takeaways

The symposium participants noted several key takeaways during the discussion, as follows:

• Land-use modelling is an important aspect of using cellular data but is inconsistent across regions or nationwide.
• A key challenge moving forward will be linking data from cellular devices to make it more meaningful.
• Validating results and determining the margin of error that one is willing to accept is critical.