With levels of homelessness at historic highs, New York City is now working to improve how it uses data in order to prevent families from entering the shelter system in the first place.
Much like how police departments around the country are now engaged in “predictive policing,” the Department of Homeless Services is now developing software that analyzes and visualizes the patterns of evictions that lead to family homelessness–and more importantly, predicts which neighborhoods, buildings, and even addresses to target its resources and outreach efforts.
“[Eviction] is a major cause of homelessness for families who come into shelters,” says Sara Zuiderveen, assistant commissioner of prevention services at DHS. “The challenge really comes in the targeting piece, making sure services are reaching the people who are most at risk.”
Among homeless families, about one in three first enter the shelter system after experiencing an eviction. But for the few hundred prevention and outreach staff at the NYC Department of Homeless Services, reaching these families before they show up at a shelter is like trying to find a needle in a haystack–only 5% of the some 200,000 eviction notices filed in New York each year cause families to become homeless.
Typically, evictions take several months from the first court filing to the point when a family gets booted out the door, so in theory that’s a vital window in which DHS and its network of partners has to act, says Zuiderveen. Between 2009 and 2013, about 4,800 families entered a homeless shelter in the months after an eviction filing against them.
DHS is working on the project with the SumAll Foundation, the charitable arm of a NY-based marketing analytics startup. Over the last few months, staff at the foundation spent time analyzing DHS files with shelter check-ins, where a family is asked to provide its most recent address, and matching these addresses to eviction court records. In that way, it could create the visualization (seen below) that highlights which evictions led to homelessness. Its next step was to use that data to find patterns and make predictions.
The model is still being tested, but SumAll Foundation CEO Stefan Heeke says he’s excited about the impact of the work. “We can confidently say that we are in a position to make that prediction,” he says. “The idea is to create an application–literally a map–where we can basically point out the hotspots or the buildings or even the addresses where eviction filings are happening and which are likely to result in homelessness.”
Right now, he says, they’re working on developing a dashboard that DHS staff will be able to use to help plan their day. That might involve social workers visiting a building and leaving information about legal services, housing programs, and other services that might keep the family in a home.
Of course, DHS can do all the predicting it wants but better funding and programs are needed if it is to be successful at stemming the rising tide of homeless families. According to a recent report by the Coalition for the Homeless, for example, former Mayor Bloomberg’s 2011 cancellation of a rent subsidy program has sent more families to shelters. New York City’s new mayor Bill de Blasio has said he will make the issue a priority of his administration, but it is still only a few weeks into his term.
Zuiderveen says the prediction algorithm could help win more funding. “Programs like this will help us show we are using our resources in the absolute best way possible to help the people need them,” she says. “Always the challenge with prevention is that you have to prove that something didn’t happen.’”