Milwaukee County experimented with machine learning to analyze statues and policies and identify biased or discriminatory language.
To identify potential inequity in the outcome of foreclosure policies and procedures, technologists in Milwaukee County, Wisconsin, turned to artificial intelligence.
The experiment was part of the 2022 SAS Hackathon, a six-week-long annual event in which teams work with experts and SAS mentors to find innovative ways to solve challenges. The county chose to use their access to SAS Viya software running on Microsoft Azure to study foreclosure data as part of a larger effort to reduce inequity.
In 2020, the County Board passed an ordinance declaring racial equity a public health crisis and committing the government to, among other initiatives, “track and analyze data to better understand the impact of County services and find solutions accordingly.”
One of the county’s projects called for analyzing regulations from “years and years and years ago” to find places that inappropriately codified racism into statutes, said Kathy Henrich, chief executive officer of the two-year-old Milwaukee Tech Hub Coalition, which has the mission of inclusively doubling tech talent in the region. Instead of manually reading through all the statutes, the team leveraged SAS machine learning to analyze the words of county policies and ordinances for biased or discriminatory language. The team used “AI technologies to help uncover most likely places so that they could then dig in further,” she said.
Henrich and her team worked with SAS and the county to define the challenge statement, and then the county collected the data to be analyzed. That amounted to reaching out to holders of publicly available foreclosure data and receiving Microsoft Excel spreadsheets that were uploaded to Viya, said Lynn Fyhrlund, chief information officer for Milwaukee County.
“Because we only had 30 days to do this process, there wasn’t much of a formalized procedure. It was, ‘Hey, where can we get the data? How fast can we get it and can we get it into our system?” Fyhrlund said. “Part of this, to be very upfront, is we realized that we have lots of data. It’s siloed all over the place and there’s not been data stewards around it. It’s something we’re working on internally ourselves, but this project highlighted how important your data is.”
Additionally, it was the county’s first foray into artificial intelligence and machine learning – areas the county doesn’t have much expertise in, he added.
That’s OK, said James Caton, head of smart cities partnerships at SAS. “Customers come to the hackathon with the problem and the data,” he said. “We provide that structured environment for them to organize the data, analyze the data, validate the insights. It’s a low-risk and high-reward way for cities and organizations to innovate because they’re able to work in a structured environment that provides all the tooling and the mentoring for them to really focus on a problem and very quickly iterate on it and come to some resolution.”
The results from the hackathon, which started March 1, were inconclusive, Fyhrlund said. “But that also tells us a lot about our data: Where is our data going to come from? How are we going to put that data together? Do we have to look at how we’re collecting data and go back to departments and processes to make sure we’re collecting the right data?” he said.
He said the experience worked in parallel with other related equity initiatives. For instance, the county has a strategy dashboard that serves as a performance management tool and supports transparency and accountability. It currently includes data on the workforce, mortality and overdoses. Future datasets include demographics, health rankings, housing and employment.
What’s more, the county recently created an enterprise data services team and last month created a data strategy. “It wasn’t because of this process,” Fyhrlund said of the hackathon. “I think what this process did was help reinforce that we’re on the right path.”
Stephanie Kanowitz is a freelance writer based in northern Virginia.
NEXT STORY: How can cities respond to extreme heat?