Economic inequality is often discussed in terms of the communities where people live or work. Khoury College alumnus Dan Calacci (BSCS ’16) is intrigued by the idea that a kind of segregation “also happens in the places where people hang out.” Referred to in sociology circles as the Third Place – not home or work, but where people spend their leisure time – Calacci’s recent work focuses on demographics in the places where “people share ideas, talk and hang out, where you can be with people but not responsible for hosting them.”

As an undergraduate, Calacci took the only ML/AI class available. (As of now, there are five or more undergraduate and graduate courses at Khoury College in that specialization). For two years, he worked with David Lazer, professor of political science and computer science, studying computational social science in an interdisciplinary space. Calacci discovered he could use computation to test and measure sociological and political theories, a realization “that was revolutionary” to him, he remarks.

His ML knowledge has grown at MIT, where he is currently a PhD student and researcher with the MIT Media Lab. He is most interested in understanding how patterns of socio-cultural diversity affect human behavior.

Atlas of Inequality Incorporates Existing Data

His interest in these patterns led him to work with Esteban Moro, a visiting professor at the MIT Media Lab and associate professor at the Universidad Carlos III de Madrid in Spain. Prof. Moro is leading their project, the Atlas of Inequality, a colorful and interactive map on the Web that highlights income homogeneity or heterogeneity of people visiting thousands of Boston locations. Unsupervised machine learning is used to cluster data points to estimate which businesses people spend time in. The map highlights places where people interact in the city outside of work and how people of different incomes do or do not encounter one another in these places. The color grid ranges from blue (very equal) to red (very unequal).

Aggregated, anonymized data for the project comes from two sources. The first source, Foursquare City Guide, commonly referred to as Foursquare, is a local search-and-discovery mobile app that accesses users’ previous browsing history to offer personalized recommendations of places to go. Cuebiq, the second source of data, is a location intelligence firm. Together, Foursquare and Cuebiq provide information regarding visitors to thousands of locations in the Boston metro area, including coffee shops, restaurants, and stores.

Calacci stresses that the team doesn’t collect data. Through Cuebiq’s Data for Good Initiative, they share anonymized location data from thousands of people. “Cuebiq gives us data,” he explains, “and we then calculate each area where people spend time.”  Since Foursquare data is mostly public, Calacci obtains information from them by “scraping the data from their website using their available API.” That data is used to create a place inequality metric that highlights how unequal the incomes of visitors to each place are.

Calacci explains, “We have results that show that places next door to each other on the same block have totally different profiles, something that we know intuitively when we live in a city. Seeing it statistically, we can use it to measure the social health of a city.”

Social Health of Cities and Urban Development

“Developers and cities often make predictions about the return on investment or credit-worthiness of a new business to measure the financial health of renting out a commercial space,” explains Calacci. What can the Atlas of Inequality offer city planners?  He responds, “We’d like to build tools to help those same folks predict and measure the social health effects of renting and development decisions.”

At the same time, Calacci is concerned that there is a risk of painting place inequality as a fundamentally bad thing. “I think it would be easy for a certain kind of person to look at our maps and say, ‘Oh, clearly, every place should be a blue dot, integrated completely across income groups.’” But Calacci doesn’t think this is always the case. “If every place in a city is completely integrated, we would probably lose our strong community places that are really important. Cafes, barber shops, bars — where the people there are a lot like you — this can be good and build solidarity, build community.”

Optimistically, he suggests that “There’s probably a balance that people in cities want to aim for, where they maintain and are part of strong local communities but at the same time have the opportunity to interact with people different than them in a meaningful way. I think that this is a really important nuance.”

Originally focused exclusively on Boston, Calacci and his team just released a new map of New York City, with ten other cities in the pipeline.

In the future, reports Calacci, they hope to work with city planners and community organizers to “explore how tools or policies could be built to leverage the data presented in the Atlas.”