Carnegie Mellon University
October 28, 2015

Predicting Human Behavior to Improve Transportation Systems

Predicting Human Behavior to Improve Transportation Systems

Assistant Professor Sean QianAssistant Professor Sean Qian

There are a number of things that can turn a morning commute into a hassle, particularly in urban areas where traffic often runs thick and sparse parking can be expensive. But with advances by Carnegie Mellon’s Mobility Data Analytics Center, a problematic commute can now be turned into a more cost- and time-efficient ride.

The Mobility Data Analytics Center, directed by CEE Assistant Professor Sean Qian, aims to make urban transit easier by helping decision makers, like state and local agencies, understand how people choose to travel.

If they can understand travel behaviors, they can determine how to best reduce congestion and emissions, such as by rerouting people around construction, changing the ways they park, or simply providing traffic information on roadway message boards.

To help agencies choose the right strategies for managing transportation systems, Qian, who also holds a joint appointment with the Heinz College, is relying on massive data.

New technologies and innovations in transportation systems have produced large amounts of data, but until now that data was scattered among different organizations that did not communicate. As a result, their data was only analyzed for a particular piece of a transportation system, such as an intersection, a stretch of freeway, or bus routes operated by the same agency.

Qian’s center aims to collect, integrate, and learn from the massive amounts of transportation data so that agencies can make smarter multimodal transportation systems. Multimodal systems include not only cars, but other modes of travel like trains, bikes, and even feet. The data that the center collects can be used to cut down on traffic in normal conditions and minimize inconveniences in unusual ones, such as when routes are impacted by construction.

“The Mobility Data Analytics Center’s mission is to track data and decide from a public agency perspective what’s the best way to manage, operate, and plan an entire transportation system network,” Qian said.

When analyzing travel behavior, Qian looks to four main areas: the modes of transportation people use, the routes they take, when they depart, and where they park. As he collects that data, he can better predict what actions people will take when their usual routes change.

Qian is currently leading a project funded by the National Science Foundation to manage parking in order to reduce congestion, emissions, and fuel consumption. He is using things like parking and traffic data, smart parking sensors, and methods for controlling traffic flow to reach the project’s goal.

Parking, while only a small part of most people’s commutes, greatly influences congestion. In fact, about 30 percent of city congestion results from drivers looking for parking spaces, according to the International Parking Institute, an association representing the parking industry and the industry’s professionals.

“Parking is a very important part of the multimodal transportation system that has been overlooked for decades,” Qian said. “If we can change parking availability, accessibility, and pricing … we can change people’s behaviors to improve the system’s performance and reduce emissions.”

For example, parking prices could be raised during congested times and lowered at less busy hours to encourage some people to travel at different times. The center can use data from parking facilities, as well as from roadways, to determine the best times to adjust those prices.

The center is also working with the city of Pittsburgh while its Greenfield Bridge is closed to test a data-based model in order to determine the best ways to manage traffic around the bridge while it is replaced. This study is one of the first of its kind in the Pittsburgh Metropolitan Area, and could validate the model to predict behavior and influence management strategies in other traffic scenarios.

As for Pittsburgh buses, the center is working with the Port Authority of Allegheny County to prevent them from bunching up on their routes, and plans to eventually use the data to improve service reliability and scheduling.

Unlike mobile applications that track traffic, like Google Maps, Apple Maps, and MapQuest, the Mobility Data Analytics Center primarily aims to serve public agencies with its data. While existing mobile applications can push data to individuals to give them the best travel options at a given moment, they do not take into account people’s behaviors upon receiving that data. If a significant number of people use those map suggestions, a roadway that was once not congested could become congested.

“Map applications provide information in real-time, but are not interested in optimizing transportation systems overall,” Qian said. “They can’t provide that type of information for our city managers and planners. If you optimize the entire system, every individual will benefit from it.”