Big Data

How one city government harnesses data to improve efficiency, save money, and protect the environment

Twice a week in Louisville, KY, different groups of government leaders get together to analyze numbers. Here's how the city is coordinating strategies around its data.

20160830113010.jpg

At a LouieStat meeting in August 2016, city leaders talked about meeting goals to reduce energy use.

Image: Hope Reese/TechRepublic

In 2012, Theresa Reno-Weber stepped into the role of chief of performance improvement in the city of Louisville, KY. It was a year after Mayor Greg Fischer—who had had a penchant for data—took office.

Fisher "wanted to bring into government the practices that he felt had made him successful as a businessman," said Reno-Weber. "Making data-informed decisions was a top priority." Here's how the city did it.

Implementing a brand-new data strategy

Reno-Weber began strategizing about how the government could make sense of its data. Fischer had "done his homework," she said, looking at how cities like Baltimore, MD had implemented data programs. "Baltimore asked, 'Where are we getting the most calls and complaints about trash? Where are we getting the most calls for service around improving our roads? How do we use that data to improve and proactively deliver services?'" Reno-Weber said.

Louisville adopted this model. At first, it wasn't about not having the data, said Reno-Weber—it was about how the data was being used. "A lot of data was put into systems but wasn't being looked at on the backend," she said. "Regulation inspectors might input data about code violations and properties, so we could ask, 'Where do I need to go and do a re-inspection?' But no one was looking at it from a standpoint of patterns or prevention."

LouieStat—the city's data strategy meetings—are about problem-solving around the data, Reno-Weber said. Currently, LouieStat forums, which include the mayor's leadership team and department heads, are held bi-weekly. The topics rotate—switching from human resources to public health to sustainability.

Departments might have previously argued that they couldn't improve in a certain area for lack of funding, or staffing, or technology. "Now, the leaders can either throw a flag on the play and say, 'No, it's not necessarily it.' Or, 'You're right. Let me provide you something,' or 'Here's why we don't have the resources,'" Reno-Weber said. Instead of having multiple meetings, issues are addressed immediately.

"It's not enough to come with instincts and anecdotes," she said. "The decisions should be rooted in data."

Collecting and reporting data

Each department uses a standard reporting form based around key performance indicator (KPI) input. The departments account for financial health and measures of safety, Reno-Weber said.. Questions include: Who is impacted? Who is unsafe? Who is spending a lot of money on overtime?

Performance coaches meet with the departments to work on tailoring the key performance indicators for their operational area, then report the most relevant data. The template can be tailored for each department.

What data matters?

"There were a lot of things that maybe a supervisor or manager cared about that didn't necessarily need to be brought before the mayor and all of these functional heads," said Reno-Weber.

The strategy is to "find metrics that spoke to the things that the mayor was trying to do," Reno-Weber said. "To increase educational attainment, or improve the health of our citizens, we want to look for metrics in our public health department that speak to health of our citizens. What's our library doing to help increase literacy rates? Who are they getting in their programs to do those things?"

Reno-Weber discovered a key to figuring out the most pressing problems. "I'd ask, 'You're the head of the jail, what is keeping you up at night? Okay, let me find data around that problem so that we can start to identify something that will help improve that,'" she said.

Another important question: "What are the decisions that you make on a daily or weekly basis that it would be helpful for you to have more data to support?" Reno-Weber said. "Cutting to the chase was really helpful."

Data success stories

Early on, the public works director said the city was "getting beat up on trash pickup." The majority of 311 calls were reporting missed trash pickups, and the director argued that the department wasn't staffed well enough to meet the demands.

SEE: Video: How the private and public sector can work together to benefit from technology (TechRepublic)

"We asked him to pull the data," Reno-Weber said. So the director looked at 311 calls. How many trash pickups happened each month? Where were the missing pickups? What time were reports called in?

The data showed 300 or 400 calls a month about missed trash pickups out of almost 800,000 opportunities—less than a .01 error rate. They saw that it wasn't a priority to fund another pickup.

Location-based data

The next phase, said Reno-Weber, is to overlay the department's location-based data.

The community services director noticed that the concentration of fires correlated with the community her department served. So they looked at where community services caseworkers were working with fires, and discovered that 25% of the fires in the last four years occurred in the homes where community services was working with the family. "We had an opportunity for community services caseworkers to be cross-trained in fire safety," Reno-Weber said, "to flag for a family if they had a smoke detector or they had some fire hazards within the home and they could educate them on it."

Recommendations for getting started with data

Don't wait for the data to be perfect, Reno-Weber said. "People are really hesitant to start to present their data." But the data gets better when it's public, she said. You don't necessarily know that the data is bad, Reno-Weber said, until you start to analyze it and ask some critical questions. The person putting in the data needs to understand how to input it in a way that's effective and consistent, so they understand the value.

Making data public also allows you to crowdsource, she said. For instance, Louisville's crime data is open to the public and geo-coded, so people can see where crime was happening. A few years ago, there was crime on the waterfront.

SEE: Video: How city government is using data to meet goals (TechRepublic)

"It was very public, all over the news," Reno-Weber said. A few reporters took a look at the site and didn't see crime showing up on the waterfront. "They said, 'You guys are trying to hide something,'" Reno-Weber said. But when the team looked into it, they discovered that the waterfront didn't have an actual address, so the system was coding the crime to a nearby garage across from a 911 center.

"We went in and fixed the location, the latitude and longitude, and it worked. We don't have a lot of resources in city government to sift through all of that," she said. "It's better to open that up to citizens and say, 'Hey, if you see anything wrong, let us know.'"

Challenges to data strategy

Louisville Metro Government doesn't have a data warehouse, and Reno-Weber said that merging systems is difficult. Also, "a lot of systems are built around keeping data locked into a box, not pushing it out to the public," she said. "Now, we're asking these systems to allow us to put everything out, and that's just not where a lot of our systems are. That's frustrating, because any time you want to do that multi-layer analysis, we have to do it manually."

Funding is also an issue. "We can't just, at a whim, update all of our technology or update all of our software," she said. "Those things have to be budgeted for. It's hard to make a case for some of those internal things, especially now when the public wants sidewalks and roads and all the other things," she said. "Saying, 'I need $15 million for a new software system,' just isn't as sexy as having your street paved."

Also, there are competing financial interests. At the end of the day, Reno-Weber said, you need to keep the city safe. "If we need more police officers, it's really hard to put a police officer up against a computer system," she said, "even though the computer system is critical."

Also see...

About Hope Reese

Hope Reese is a Staff Writer for TechRepublic. She covers the intersection of technology and society, examining the people and ideas that transform how we live today.

Editor's Picks

Free Newsletters, In your Inbox