Smart cities through data science

Friday, February 17, 2017

Bill Howe’s enthusiasm for data is matched only by his desire to improve the health and vibrancy of modern cities.

At the intersection of those twin passions is where Howe, an associate professor at the Information School, has launched UrbAnalytics, a partnership of research projects around the University of Washington campus with the common threads of data and urban issues. UrbAnalytics combines aspects of two other projects in which Howe is involved – the eScience Institute and Urban@UW.

“The eScience Institute is kind of ‘all things data’ across scientific fields, but not necessarily urban, and Urban@UW is ‘all things urban’ but not necessarily data,” Howe said. “It made sense to have a unit that’s focused on applied data science in this urban context.”

With Seattle and other cities increasingly sharing their datasets with the public, Howe saw an increasing need for a resource on campus to advance data-driven analytics and research on urban issues. A $50,000 grant from Microsoft, along with some earlier funding from the MacArthur Foundation is helping to get the UrbAnalytics project off the ground.

“As we talked to sponsors and stakeholders within the city about data projects, it became clear that we needed to provide a campus unit to talk to that was focused on doing this kind of work,” Howe said. “As opposed to a set of individual faculty members who need to stay focused on their own ongoing research.”

Working in an urban setting means working with people, who need both the tools to navigate large datasets and the ability to recognize the biases inherent in the data. If you’re working with Seattle’s ORCA card data, for example, you might need to account for underrepresentation of people from poorer neighborhoods or overrepresentation from major employers that give the cards to all of their workers. Not doing so could lead to faulty assumptions about how people use transit. UrbAnalytics will work to develop software to help people recognize such problems.

Howe advocates for the practice of responsible data science – using datasets ethically and getting to the unbiased truth that lies within them – and for research with impact.

“The first 10 years of data science was really about what we can do with large, noisy and heterogeneous datasets,” he said. “The next 10 years are going to be about what we should do with these datasets.”

UrbAnalytics will engage with city and community leaders to translate its work into action and influence public policy. To do so, the project will lean on the shared expertise of scholars, government and community leaders. UrbAnalytics’ UW partners include the eScience Institute, Urban@UW, the Taskar Center for Accessible Technology and the Center for Studies in Demography and Ecology. Beyond campus, contributors include the University of Chicago, Drexel University, the city of Seattle, and the Washington State Transportation Center.

Thaisa Way, executive director of Urban@UW and a professor in the College of Built Environments, said UrbAnalytics will fulfill an important role as a place to work with, interpret and analyze data.

“It’s not just bringing data in, but people who can think about data in complex ways and help people think about how they use the data,” she said. “We too often think about data science as just a toolbox, as something we just put in a machine and pull out. Bill can help put together those teams that can bring the data science, technology and data responsibility and work with civic leaders to use it to make better decisions.”

When people think of the “smart cities” of the future, they often envision a hyperefficient, data-driven society, but Howe sees a smart city as one that also delivers the right services to the people who need them. If data can show what services are most effective in leading homeless people to permanent housing, that’s the kind of “smart” that UrbAnalytics will focus on.

“Homelessness is a powerful example,” Way said. “We want to know what their daily routine is and how we can best meet their needs. We want to be careful about privacy and yet we also want to be able to help, so we want to think of responsible ways to collect data.”

That’s right in line with the iSchool’s strategy of making data work for the social good, and with the needs of the UW and local leaders, Howe said.

“The city of Seattle and the president of the university have been vocal that both with urban and with data, that they want the human side of smart cities. UrbAnalytics will hopefully be the center on campus to explore this,” he said.