Turning Sensor Data into Decisions in Chicago
Last year, in an ornate downtown Chicago ballroom, the seeds were planted for a new multidisciplinary research network with an ambitious purpose: to understand and improve cities. By mixing together experts in computer science, public health, education, architecture, urban planning, art and social science, the Urban Sciences Research Coordination Network (USRCN) hoped to create versatile and knowledgeable teams that could find new approaches to study cities in a rapidly urbanizing world. Sixteen months later, the early fruits of those new collaborations helped inspire a new wave of discipline-crossing partnerships at the 2nd USRCN meeting, organized by the Urban Center for Computation and Data and held inside the world famous Art Institute of Chicago.
The artistic setting befit a scientific workshop that sought to go beyond the numbers, grafting the growing potential of data analytics and computation to the deep questions that social scientists and urbanists have asked about cities and their residents for decades. That tone was set through the keynote talk by Mario Small, the outgoing Dean of Social Sciences at the University of Chicago, soon to be Grafstein Family Professor of Sociology at Harvard University.
Small discussed the study of poverty and urban density, a subject that has fascinated sociologists since the work of William Julius Wilson at the University of Chicago in the 70’s and 80’s. Based on observations of the South Side of Chicago, Wilson concluded that the consequences of being poor are compounded by the concentration of poverty, which increased after the loss of manufacturing jobs and flight of the middle class from many urban neighborhoods. When Small came to Chicago, he was interested in exploring Wilson’s theory further — not top-down using “big” data, but from the ground-up with street-level observations.
Small’s studies found that poor neighborhoods in Chicago are high in poverty but also sparsely populated with both people and resources. Compared with equally poor but far more dense neighborhoods such as Central Harlem in New York City, a more spread out Chicago area such as Woodlawn shows much higher rates of violent crime and lower measures of well-being, suggesting that not all urban manifestations of poverty are the same. Small is now working with national data on cities, as well as more fieldwork, to develop a more nuanced understanding of how different conditions affect the experience of living in poverty around the country.
“Data can dispel stereotypes, but it only works if the use of data is informed by fieldwork, and serious scientific inference versus the search for a pretty picture,” Small said.
The USRCN projects presented in a series of short talks reflected this mix of the firsthand and the numerical. In two projects, the computational method of agent-based modeling will be applied to questions of economics and health care, using deep wells of public sector and electronic medical record data to create “in silico laboratories” where the complex ripples of interventions can be tested and observed. For example, an evaluation of the HealtheRx project led by Stacy Tessler Lindau of University of Chicago Medicine will help investigators measure downstream effects of connecting patients with healthy resources in their community that a more expensive real-world trial might miss.
Other projects seek to unlock new insights from within data itself. The University of Chicago Pathways to Adulthood Project Data Enclave (UCPADE), presented at the workshop by Bob Goerge of Chapin Hall and the Computation Institute, will work with more than 20 years of data from Chicago Public Schools, covering the records of over one million students. The database will be used to look at factors affecting future outcomes of students, from high school graduation and college enrollment to employment and incarceration.
CPS is also a project partner on this year’s edition of the Data Science for Social Good (DSSG) summer fellowship, which works with data from government agencies and non-profits. Over 12 weeks, 48 fellows in Chicago will develop new tools and analyses that help these organizations make the most of their data, solving problems in energy, urban development, education, health care, and homelessness.
“A lot of non-profits are excited to do this kind of work, but don’t have the right people, and even if they had the budget, they can’t hire people at this scale,” said Rayid Ghani, director of DSSG. “Our goal was to create a workforce that’s not only trained, but passionate about doing this work.”
Other projects descended from last year’s USRCN meeting make it easier for researchers and non-researchers to work with city data. Plenar.io, created by Derek Eder of DataMade and the Urban Center for Computation and Data, is a tool for easily visualizing the treasure trove of time/space data in the Chicago Data Portal. Users can simply draw boundaries around a section of Chicago and pick the data they want to see, instead of wrangling with .csv files and mapping software. Research using these constantly-evolving datasets will drive the future of urban science, said Luc Anselin from Arizona State University, allowing for the creation of “digital neighborhoods” providing real-time information that can be used to study policy implications and urban socioeconomics.
One immediate realization of that vision is the UrbanCCD’s new Array of Things project, presented by Douglas Pancoast of the School of the Art Institute of Chicago. By placing sensor nodes around the city, the Array of Things will not only collect environment and infrastructure data from Chicago at an unprecedented scale, it will publicly release that data immediately so that users can build new applications and perform their own analytics to better understand the life of the city.
Those project demonstrations were only the inspirational first half of the USRCN agenda, with the afternoon dedicated to brainstorming new ideas. Breakout groups made up of unusual mixes of participants from academia, industry, government, and the community chewed over ideas in Education, Wellbeing, and Health, Access and Mobility, Data & Tools, Neighborhood and Community Vitality, and Natural Resources and Energy, and presented research ideas ranging from a study of reusable city infrastructure to an “open metadata portal” that would make city data more easily searchable. Those concepts might just lay the groundwork for projects presented at next year’s USRCN meeting, further seeding the growing ecosystem of data-informed urban research in Chicago and beyond.