News, Science & Research

Data research aims to predict protests

New grant from Data Science Initiative funds collaborative research using machine learning

By
Senior Staff Writer
Friday, February 15, 2019

While financial indices may seem just like numbers and a Google search just like a trivial collection of words, the potential application of both to anticipate future protests worldwide is currently under study at the University, according to the Data Science Initiative website.

A joint project titled “A Quantitative Measure of Freedom of Assembly,”  led by researchers at the University, the University of Michigan, the University of Calgary and the University of Chicago hopes to explore the application of data to predicting protests. The group was the first to receive funding through a grant called Data Science @ Brown, which is offered by the DSI.

Combining their respective expertise in economics — specifically its relation to political institutions,  governmental development and protests — the team intends to apply the latest data and economic theories to compare freedom of assembly across various nations, wrote Jesse Shapiro, professor of economics and co-principal investigator of the project, in an email to The Herald. The project will also rely on machine learning techniques, according to the DSI website. The other principal investigators are Yusuf Neggers, assistant professor at the University of Michigan’s Ford School of Public Policy, and Mehdi Shadmehr, associate professor of economics at the University of Calgary and visiting associate professor at the University of Chicago’s Harris School of Public Policy.  “We hope that our research will contribute to a better understanding of the variation in and determinants of the freedom of assembly,” Shapiro wrote in an email to The Herald, adding that the researchers are currently and “actively” using the grant to obtain and analyze data.

Danilo Freire, a postdoctoral research associate at the University, is not involved in the project, but has focused on political instability in his research. This topic piqued his interest because of his time living in Brazil, when he was exposed to the possibility of impending violence, he said. Similar to the project funded by the DSI, Freire has also engaged in research involving machine learning. His work pertained specifically to trying to predict and thereby create a warning system for genocides, he said. Timing is crucial for interventions in such atrocities, he said, adding that, “I think the question’s not to predict everything correctly, it’s just to be better than human judgment.”

Social scientists usually examine cause and effect to reach conclusions and create predictions; the methods employed in some social science research, including the protest data project and Freire’s research,hope to provide a new and potentially more accurate  way to reach the same end result, Freire said. Machine learning allows the process of social science analysis to become more flexible because it need not rely on fixed assumptions used in older statistical research methods, he added.

It is “important that someone funds … early warning systems … not only for science, but for human kind,” Freire said.

The group’s proposal was the first application received by the DSI, and it was then chosen to receive the funds. The proposal met every criteria for the grant, which includes establishing cross-disciplinary research and helping to fulfill the goals of the DSI, wrote Associate Director of the Data Science Initiative Alden Bumstead.

The DSI seeks to promote research within data science, which involves the creation and use of methods and tools to draw conclusions from data. It also seeks to increase learning within the field and look into the societal and cultural applications of data science methods. The DSI established the grant with these goals in mind, Bumstead wrote in an email to The Herald. Bumstead hopes this funding — potentially in combination with other financial assistance — will serve as a catalyst for published findings or additional inquiries within this project, she wrote.

These types of studies require heavy computer and technology use, so financial assistance from grants like the Data Science @ Brown grant further aid researchers by allowing them to take risks in hopes of achieving greater results, Freire said. There are currently more opportunities available within data science than people working to pursue them, he added; he hopes the funding of this project will increase interest in the field of data science.

The DSI hopes to award additional grants over the course of the semester, Bumstead wrote in an email to The Herald.