Mapping New York's Traffic Crashes
The New York Police Department doesn't make it easy to crunch the city's traffic fatality and injury numbers, releasing the data in a PDF format that's difficult for developers to use. A group called betaNYC has been pushing the cops to make their data machine-readable, and new mayor Bill de Blasio's Vision Zero plan might help their case.
In the meantime, it's up to some hard-working data nerds to liberate the information locked inside the files. One civic-minded hacker named John Krauss has been scraping the data from the monthly PDF releases and making it available to anyone on the NYPD Crash Data Band-Aid. Krauss runs a site called NYC Crashmapper, where you can play around with a map that displays crashes by date, what type of user was involved, and severity of crash.
The numbers he's working with have some limitations – crashes are mapped to the nearest intersection by the NYPD, so it's unclear exactly where on a block they've taken place, for instance – but it's a start.
A few others have used Krauss's data for their own visualizations. One is Ben Wellington, a visiting assistant professor at Pratt Institute's Grad Center for Planning who teaches urban planning students about statistics by having them look at real open data about New York.
"I made the whole course around open data for people who are actually studying the city," says Wellington. "It's very applied. My goal is to take people that aren't very interested in statistics and make it interesting."
Wellington recently started a blog called I Quant NY, devoted to quantitative analysis of New York data, and he's gotten a huge response to the maps he's created with the Crash Data Band-Aid numbers. One map shows how certain neighborhoods, including Williamsburg, Brooklyn; Queens Boulevard; and the Grand Concourse in the Bronx, are hotspots for traffic fatalities. Wellington’s analysis shows that 23 percent of New York's 2013 traffic-related fatalities occurred in just 5 percent of its neighborhoods He also did a map that shows all 2013 bike crashes that resulted in injuries reported to the police, a 3,800-dot visualization that he, a regular bike commuter, calls "terrifying."
2013 bike crashes that resulted in injuries reported to the police.
Being the numbers guy that he is, though, Wellington doesn't draw hasty conclusions from the way the data looks. Factors such as density and traffic patterns all need to be accounted for. "You have to be careful about looking at a map with colors," he says. "I do my best. If I make a conclusion, I try to make it clear that it's a hypothesis." And he doesn’t necessarily want to get fixated in the most sensationalist aspect of the visualizations, those hot spots. Fully 60 percent of the city's 2013 traffic fatalities took place in neighborhoods with two or fewer deaths. And looking at "bright spots" where there are no deaths might yield other important insights about how to make New Yorkers safer on the street.
A heat map of 2013's traffic fatalities.
Wellington has been struck by the intensity of demand for a thoughtful treatment of this data in a city with a growing awareness of traffic safety as a public health issue. He also understands why the police might be wary about putting detailed information in the public realm, even if he thinks they should.
"Nobody likes being watched," he says. "It can be a little scary. They have a point, people can use data to tell any story they want." But as a statistician, he believes that opening up government data allows the most reasonable conclusions to rise to the top, ultimately benefiting the public.
He acknowledges the importance of the newly prominent emotion-based campaign to change the city’s traffic culture, which is being led by family members of victims. But emotion alone won’t be enough to get bureaucrats and elected officials to make life-saving changes, Wellington says.
"People have the feeling that the city is unjust in some ways, but feelings don't drive policy," he says. "Sometimes you need data. When people try to change policy, they use data."