For yeaRS, Boston’s Department of Transportation has collected GPS data on every taxi pickup and drop-off throughout the city. It is an astonishing accumulation of raw numbers on how Bostonians get around, ripe with opportunity for analysis.
Until recently, there was a problem: There was no one to dive into it. “It was basically sitting on a hard drive waiting to be looked at,” said Vineet Gupta, director of policy and planning at the Boston Transportation Department.
Last November, the city handed over a big chunk of that data—comprehensive cab data from May to December 2012, or 2.3 million taxi rides—to the Massachusetts Institute of Technology. In a competition that ended in February, MIT offered two $4,000 prizes for the student groups that came up with the best interactive visualization and the best prediction model for taxi demand at spots around the city.
The winning maps provide a sweeping look at the taxi pickups and drop-offs throughout an average week, a kind of MRI of the patterns of human activity in Boston. They also give a sense of what it takes to fulfill the promises offered by big data—and the hurdles that lie between collecting vast quantities of information and actually improving the system.
Gupta hoped that students would help the city answer a basic question, he said: “Are cabs getting to where people want them?” But he also wanted to see trends that planners simply didn’t know to look for, and that even the savviest cabbie might be unaware of as well.
In the visualization that took first place, designed by electrical engineering and computer science PhD student Gartheeban Ganeshapillai, a timeline of a week’s worth of pickups plays like a short movie. The regular patterns in Boston’s taxi demand thrum like a heartbeat on a heat map; the pulse grows stronger and bright red as demand hits a fever pitch during weekday rush hours and on Friday and Saturday nights.
In many of the visual presentations, a recognizable portrait of Boston emerges. A 24-hour city, we are not: Demand for taxis takes a nosedive after 10 p.m. on most weekdays and 1 a.m. on weekends. Runmin Xu, who is pursuing master’s degrees in environmental engineering and transportation as well as in electrical engineering and computer science, pointed out a way to spot Boston’s entryways from points beyond: On weekday mornings, pickups are spread relatively evenly throughout the heart of the city, except for three distinct spots of heavy demand at Logan Airport, North Station, and South Station.
Other trends are less intuitive. Residents’ tastes in nightlife skew eclectic on Friday nights, with taxi pickups at the end of the night dispersed more in the South End and the Seaport District. On Saturday nights, it seems, revelers head in greater concentration to the city’s tried-and-true late-night haunts: the areas around Faneuil Hall, Fenway Park, and the Prudential Center.
Rain causes a spike in demand for taxis—and that demand appears to stay strong several hours after the end of the downpour, suggesting that passengers are more likely to opt for a cab if they see puddles on the ground and fear it may rain again. But pickups and drop-offs stayed at a near constant when the temperatures dipped low, said engineering graduate student Yingxiang Yang, one of the winners of the prediction challenge. Hardened New Englanders don’t like to get wet, apparently, but they will fearlessly brave the cold.
Farther afield from the heart of the city, spots such as the Forest Hills and Fields Corner T stops are largely quiet throughout the day—except for a surge in demand each weekday at 8 a.m. That suggests that people are using taxis to complete their morning commute in neighborhoods inadequately served by the T, according to Ganeshapillai. “People take the T to get close to the place they want to go to, and for the final leg of their journey they take a taxi—rather than take a taxi all the way,” he said.
Some of the trends, the competition participants said, were downright infuriating, and suggest that Boston’s public transportation system could stand to improve its outreach to visitors. One of the most common destinations for taxi pickups at South Station is the airport—even though the MBTA’s Silver Line makes the exact same trip, at a fraction of the cost. Similarly, one of the most popular destinations from the North End is the New England Aquarium, which, Ganeshapillai pointed out, is about a 10-minute walk away: “A taxi ride would take you longer!”
The notion of using data to improve transit offerings is an idea that has taken on momentum in Boston, and throughout the country. New York and San Francisco have released similar taxi data to eager wonks willing to perform the heavy lifting of analyzing millions of GPS points and time stamps. Hubway, Boston’s bike-share program, performs an annual data dump of all its ridership records. The Metropolitan Area Planning Council and the Massachusetts Department of Transportation just announced a competition to analyze the state’s vehicle census, a huge compilation of data that includes the age, model, estimated mileage, fuel efficiency, and ZIP code of all the cars and trucks registered in the state between 2008 and 2011.
Jameson Toole, an MIT PhD student in engineering systems who was among the winners for the data visualization, said he envisions large data sets becoming a key tool for taxi companies. “I think everyone has experienced trying to get a cab in downtown Boston on Saturday night at 1:30,” Toole said. “That’s an indication that things are not moving as smoothly as they could.”
And if city cab services would rather put off the burden of number crunching, they now have another reason to get serious: competition. Black car services and ride-share program such as Uber and Lyft might be controversial for their surge pricing and circumvention of the city’s taxi medallion system, Toole said, but they are leading the way in focusing on hard data about customer need. “You can see that they are finding ways to improve their services by digging down in this data and preemptively placing drivers in places,” Toole said, “and that’s good for both drivers and riders.”
At the Boston Transportation Department, city planners have only begun to take a closer look at the trends and patterns illustrated by these visualizations. “We’re just beginning to have the conversation relative to how we can use this data to inform new policies—for example, relocating cab stands,” Gupta said. But big-data challenges like this one could be an important step toward planning a city where residents will finally be able to find a cab when they want one.