Jul 16 2014

Applying Analytics Can Improve Outcomes

Everything from building safety to child welfare can benefit from the use of Big Data.

Several years ago, New York City experienced an unacceptably high number of fires in illegally converted apartment buildings. In 2005, two firefighters died while attempting to rescue residents in a blaze that injured 20. City officials knew older buildings with limited exits were the primary source of the problem, but when they began to explore the data more closely, they discovered an interesting pattern.

In addition to being built before tighter regulations were enacted in 1938, buildings that also had ongoing foreclosure proceedings, active tax liens and complaints were more likely than other illegal conversions to go up in smoke, says Nicholas O’Brien, acting director of the Mayor’s Office of Data Analytics. MODA analysts of the Bloomberg administration pored through data to target older buildings whose owners were in financial distress. Those buildings were placed at the top of the list for inspection and likely closure within 48 hours instead of the typical 40 days, saving lives and preventing fires.

Call it Big Data meets the Big Apple. In the years since this experiment, New York City has enlarged its data-driven government initiative, using a data warehouse fed by analytics to make life better for its citizens in myriad ways. The idea is to use analytical tools and techniques against data from a number of city agencies to discover trends and ideally, take action before something bad happens.

Throughout the public sector, the use of analytics helps governing bodies to improve the health and safety of their constituents. And those cities and states that aren’t yet leveraging Big Data are talking about it. Different technologies can be used to identify trends — from the humble spreadsheet to advanced analytics platforms with many options in between. The possibilities are endless, and the potential to improve lives is seemingly without limit.

Focusing on Results

The New York MODA relies on some powerful tools, including an Oracle data warehouse and an analytics platform. “We’re seeing huge growth in our use of analytics,” says O’Brien. Rather than focusing on an activity (such as inspecting older buildings, for example), the team now focuses on an outcome — fire prevention in this case.

O’Brien heads a team of eight, including programmers, econometricians and business analysts who understand the data and the applicable regulations. Much of the analytical work is done on mainframes, but there are parallel databases just for reporting. “When we query this data, we can’t have it interfere with production. So we have a separate decision layer,” O’Brien says.

With a few years of analytics experience in hand, NYC recently set its sights directly on human safety. “We are not as far along on that because there are privacy concerns and federal protections on the data that prohibit us from using it to do cross-agency initiatives,” O’Brien says. “We have to go a little more slowly and deliberately.”

Maryland has had considerable success in leveraging data to improve the health and safety of its citizens. By applying resources where they are most needed and in a targeted fashion, the state reduced its infant mortality rate to 6.3 per 1,000 live births — a drop of about 21 percent since 2008.

Another key area of focus: Improving the welfare of children within state custody. Maryland StateStat director Matt Power says that a few years back, the director of the Department of Juvenile Services (DJS) began asking some very pointed questions, such as, “How many kids are in our care right now? What is the average stay in our care? How do young adults fare when they age out of the system?”

“There weren’t great answers to those questions,” says Power. Supported by Maryland Governor Martin O’Malley and StateStat, the DJS quickly adopted a data-driven approach to problem solving. “This has been a game changer. You can look at length of time in care and lots of other metrics — we have so much data they can use to manage. It changes the way we look at what we do. The numbers have moved to exactly where we want them,” he says.

In 2007, 50 youths under the age of 17 were murdered, with 22 of them under DJS supervision at the time. By contrast, in 2013, 20 were killed, and only two of them were under DJS supervision. That’s two too many, as Power readily acknowledges, but it’s a major improvement nonetheless. The lower murder rate is partly attributable to the use of analytics as well as more strategic deployment of police.

Focusing on Families

A relative newcomer to analytics, Colorado recently implemented a portal for child protection and child welfare data across the state and counties. “This is an effort to use data and outcomes to improve policy and practice across the state of Colorado for child welfare,” says Jonathan Sushinsky, manager of research analysis and data for the Colorado Department of Human Services (CDHS).

“We felt we should measure how we were doing with our population of children,” says Sushinsky. Colorado added measures to gauge progress, such as a scorecard that tracks metrics related to goals, by county. The scorecard pulls data from a large transactional database via an analytics reporting program developed by the University of Kansas School of Social Work.

The tool enables county child welfare professionals to view their outcomes and compare them with other jurisdictions. The goal is continuous quality improvement through the sharing of best practices, adds Katie Facchinello, communications specialist for the CDHS Office of Children, Youth and Families. The public can access top-level reports as well, which enlists their help in keeping children safe, and also increases transparency.

These data-driven projects are especially rewarding to take part in, says New York City’s O’Brien. “We’ve seen the number of fires go down, and that’s how we know we’re doing better,” he says. “It’s a great job because we get to deal with a lot of different problems. And there’s so much more we will do.”

Tragedies Spur National Interest in Analytics

Many states are stuck in a comparative stone age, at least when it comes to technology used to improve child welfare. In Massachusetts, for example, estimates of the number of children missing from the Department of Children and Families numbers in the hundreds, according to The Boston Globe. Too often, the missing are later found dead.

The Florida Department of Children and Families, too, is in crisis, with 477 child deaths since 2008 within families that had been flagged as dangerous, according to the Tampa Bay Times.

As a result of public outrage surrounding these tragedies, Massachusetts and Florida have joined other states, including Indiana, Pennsylvania and Texas, in implementing — or at least exploring — tools that will allow them to better predict when a child is in danger, whether in or out of state care. These states are exploring web portals along with robust analytics tools that can better gauge child welfare.

Andrew Kist