Data-driven transit advocacy, with a side of cloud

A National Science Foundation-supported program seeks to ease the use of Baltimore’s public transit system for low-income residents.

With a grant from NSF and additional support from CloudBank, another NSF-funded initiative, Vanessa Frias-Martinez, an associate professor at the University of Maryland, is working to put data collection in the hands of residents to enable self-advocacy.

In response to longtime complaints about late buses and other transportation problems, Frias-Martinez “proposed to … build a toolkit, which is basically a mobile app and a dashboard,” she said. “The mobile app will allow residents to collect data about their door-to-door mobility patterns, so it’s kind of a digital diary.”

When residents leave home to use public transit, they start collecting data about their trip on the phone. During the journey, the app asks about their perception of the quality of the service they’re receiving. At the same time, Frias-Martinez’s team collects real-time information from the Maryland Transit Administration (MTA) and data on how long the trip took. Using artificial intelligence algorithms, all the data is analyzed and presented on a dashboard.

The goal is for participants to use the data to support their claims in meetings with MTA and Baltimore City Department of Transportation representatives. Ultimately, that data-driven advocacy could effect changes to the transit system, said Frias-Martinez, who leads the university’s Urban Computing Lab.

“We want to collect the data that will allow them to back these types of assertions” about lateness or routes that don’t go where jobs are, she said. “If we have data from multiple residents from multiple neighborhoods in Baltimore City, and they can show with the data that that’s what’s happening, we felt that would give them a stronger argument in those conversations.”

Ultimately, Frias-Martinez said she hopes to have a toolkit that other cities and even state governments could adopt.

“The main outcome of this project will be to test whether these types of tools are actually useful for what we think they might be useful for, which is collecting data that will allow neighbors, residents to back their advocacy efforts,” she said. “In our proposal, we said that the software we’re going to design was going to be open source, then other cities and other departments of transportation could use it.”

Her team is partnering with five local university researchers and two organizations to identify participants. One is the Housing Authority of Baltimore City, which oversees public housing, and the other is the Central Maryland Transport Alliance, a public transit advocacy group.

They’re also working with CloudBank, a facilitation service to help researchers access and use public clouds. It went into production in August, about two months before Frias-Martinez received her grant. Both are four-year initiatives.

“The primary purpose of CloudBank is to support researchers who get NSF awards through other solicitations that have requested cloud resources,” said Mike Norman, director of the San Diego Supercomputer Center and a physics professor at the University of California-San Diego.

Modeled after a commercial bank, CloudBank holds grant funds on accounts and pays them to cloud providers based on customers’ usage. When researchers win an NSF grant that requires cloud access, the funds requested for cloud resources go to CloudBank, which works with Strategic Blue, a U.K.-based company that helps organizations procure cloud services. Strategic Blue pays the cloud providers for usage, and researchers can see their rates at the CloudBank online portal.

Currently, only NSF Computing and Information Science and Engineering directorate solicitations may use CloudBank. Frias-Martinez is the first -- and so far only -- researcher taking this route. Norman said he expects more projects to take advantage of CloudBank when NSF makes more awards at the end of the fiscal year.

Another way into CloudBank is through CloudBank Enterprise, Norman added. It facilitates a conversation between an institution and Strategic Blue about cloud infrastructure onboarding.

A third aspect is CloudBank for the classroom. Instructors fill out a form on the CloudBank website and ask to have a data hub spun up in the cloud for their courses. Students and instructors can click a link, and all course material will be pulled into CloudBank. About six customers use CloudBank this way now.

“The idea is that these universities who may or may not have current relationships with the cloud providers, would like to take advantage of our financial engineering,” Norman said. “By and large, scientists don’t use the cloud for their computations. They use resources under their desk or they use campus resources. There’s an aspect of this project of evangelizing the value of the cloud and getting adoption.”

Frias-Martinez said her project will use CloudBank for its duration. Benefits include having one point of communication for all cloud needs, including understanding or disputing charges or engineering help.

“I think in the long run, we will also save money,” she added, because for some of the services that the project is running, CloudBank has suggest less expensive options. CloudBank works with all of the major cloud providers: Microsoft Azure, Google Cloud and Amazon Web Services, which is what the Baltimore project uses.

She hopes other researchers take advantage of CloudBank. “Working with cloud services is a lot of work. It has a really steep learning curve, and my hypothesis is that a lot of researchers try not to use it because it’s hard to learn and use,” she said. “If they knew there’s CloudBank to provide this type of support, I think more people would be encouraged to develop technologies together with them. It might be a matter of not knowing that the service is there.”