The Bureau of Transportation Statistics generally operates at a slow, meticulous pace while it collects data on topics such as on-time airline departures, maritime activities and state transportation systems — then assuring and reassuring the data’s accuracy before releasing it.
Now, with the U.S. economy having just experienced its worst financial quarter in a decade because of the coronavirus pandemic, senior government leaders are asking when it will recover. The Transportation Department’s statistical agency may be able to offer clues.
This new mission means BTS has to release transit data daily, a paradigm shift for an agency that typically takes months to release numbers. The 60-person bureau is now using new investments in information technology and artificial intelligence to provide daily transit statistics across more than 40 metrics to government leaders, including Transportation Secretary Elaine Chao, and other senior officials involved in the short- and long-term strategy planning for economic recovery.
“Transportation is almost like a precursor, if you will, of how the economy will recover,” Patricia Hu, director of BTS, said in an exclusive interview with Federal Times.
Consider freight shipping. Hu said the bureau’s data shows that freight movement serves as an early indicator of economic health by three to four months.
“If we start seeing the freight movement starting to pick up, we know that the economy is going to pick up soon,” she said.
But to help track economic recovery, the agency must move significantly faster. For BTS, this has required a complete shift of mindset, rapidly transitioning from a focus on statistical quality to timeliness of the data.
“The new mindset is: ‘Let’s send out a flash indicator right now — we can refine that later on,'” Hu said.
Emerging tech as a helpful hand
BTS recently adopted flash indicators, or data that provides a snapshot into specific metrics that the agency uses to monitor the immediate impact of COVID-19 on the transportation system, Hu said. The bureau also now releases preliminary data — something it didn’t do before the pandemic.
For example, Hu said, if 95 percent of airlines release passenger data to BTS, then it immediately publishes that data without the typical intense interrogation for statistical quality. It later reviews the information before publishing the final results.
BTS is integrating data from multiple sources, such as mobile phones, subway system numbers, and passenger screening data from the Transportation Security Administration. BTS is monitoring 43 indicators — covering passenger and freight transportation demand across the air, water and land transportation — to track the trajectory of the U.S. economy.
To speed up data collection, there’s often a reliance on “proxy data,” which can serve as a strong indicator for travel trends. For example, BTS analysts consider the TSA’s reported number of passenger security screenings at airports a “pretty good” way to gauge air travel, Hu said.
The COVID-19 response is a unique challenge for the statistical agency. While the agency has experience tackling a hurricane or earthquake response, the pandemic response lacks what natural disaster responses have: historical patterns. So when BTS starts to observe an uptick in public transportation ridership in Washington, D.C., how that fits into the recovery forecast is unclear because there’s no historical context.
To solve that problem, the agency is “exploring” artificial intelligence solutions, Hu said. The bureau developed a data quality tool that uses natural language processing to pre-process transit data and prepare it for quick data analysis.
BTS also created a data collection tool using Python code that automatically harvests web data from large transit agencies across the country to collect data it needs. It has created another model to track turnstile counts in other transit systems to seek out anomalies.
“We want to see at what point we start to see upticks and do we see the upticks in a consistent manner ... because we don't want to prematurely indicate that the economy is under recovery,” Hu said.
In March, BTS established a “statistical special force” as the new coronavirus continued to worsen throughout the United States, Hu said. To ensure the bureau was still using its resources effectively at this more rapid pace, BTS officials, including top statisticians, economists, geographers and transportation specialists, hypothesized how the virus would impact how people and freight move. Based on these hypotheses, BTS decided what data it needed to collect in order to assist policymakers.
Data curated by BTS has also played a critical role in the coronavirus relief effort. Hu said her bureau’s biggest success was that its data was used to disperse $50 billion in funds to airlines as part of the government’s recent coronavirus relief package.
The BTS data, Hu said, was able to expedite how quickly the Treasury Department distributed the money. That experience itself highlights how the pandemic has changed the bureau’s role.
“In this expedited schedule, we don’t have time just to kind of play around to see what information’s out there,” Hu said.
Andrew Eversden covers all things defense technology for C4ISRNET. He previously reported on federal IT and cybersecurity for Federal Times and Fifth Domain, and worked as a congressional reporting fellow for the Texas Tribune. He was also a Washington intern for the Durango Herald. Andrew is a graduate of American University.