The Penn Program on Opinion Research and Election Studies hosted its first-ever virtual hackathon earlier this month, tapping both undergraduate and graduate students to tackle real-world, real-time datasets about the coronavirus crisis. The groups were given just two bits of advice: Be creative and discover something interesting.
About 20 students from a range of disciplines including engineering, city planning, public policy, political science, and education formed nine teams and were given datasets. They had the school week to dig into the data and lay out their findings before presenting their findings to a panel of faculty judges from PORES in a Zoom meeting.
“One of the things that I always tell my students is the best way to get better at data science or statistics is to be motivated by a project that you’re interested in learning about,” says Stephen Pettigrew, associate director of data sciences at PORES and the co-organizer of the hackathon. “I think right now, everybody is really interested in understanding more about what’s going on during this pandemic, so that was part of the idea for the event.”
The teams were given raw data from publicly available sets, including a number of public health datasets like the database kept by Johns Hopkins University, as well as projects that news media sites have launched tracking cases and deaths, public opinion polls, and different surveys that have been run in the past month that asked questions regarding the outbreak.
“A hackathon is a perfect event to host in a time like this because it’s something that can be done entirely online very easily. We figured we could capture this moment and have the hackathon focus on coronavirus since there’s lots of cool data out there,” Pettigrew says.
Having a diverse group of students from varied disciplines working together added to the quality of the findings, says Samantha Sangenito, a data scientist who teaches at the Fels Institute of Government and co-organizer of the hackathon.
“You’re getting so many perspectives because all of these students come from different programs where they’re really learning different things,” she says. “So when they come together, they all learn from each other and really capitalize on that knowledge.”
Unlike a traditional hackathon that has one winner, the PORES event offered three winning categories: Best Visuals, Most creative, and Most Interesting Finding.
The teams had about five minutes to present their findings to the judges, which consisted of Sangenito and political science professors John Lapinski, Matthew Levendusky, and Marc Meredith. The judges then gave feedback to each team.
Haley Suh and Grayson Peters teamed up to look at social distancing efforts by demographic, which won the Most Creative category.
“I think our finding was surprising for a lot of people, including us,” says Suh, a senior from Chicago majoring in political science with a minor in PORES.
She and Peters found that people who identify as conservative were 10% more likely to engage in social distancing measures than those who identify as politically liberal and people living in rural and suburban areas were more likely to social distance than those in urban areas. Social distancing was also found consistent across income levels, which Peters says ran counter to what they expected to find.
“Working with this real-world data in real time allowed us to pull conclusions that were salient to our daily lives right now when we’re at home doing these social distancing measures,” says Peters, a sophomore from Nashville majoring in environmental studies, who also does curriculum development for PORES. “The hackathon provided a great real-world, hands-on opportunity.”
Sangenito says it was difficult to pick the winners because all of the findings were interesting and well done. They’re now considering holding another hackathon, perhaps in the summer or early in the fall semester.
“It really gave the students from across the University a unique opportunity to collaborate with each other” Sangenito says. “It showed everyone that this is what the data sciences at Penn look like: cross-disciplinary, practical, accessible and exceptional. They were just so impressive.”
List of winners:
Best Visuals—“Stress by Demographic,” Leo Chen, Caroline Riise, Annika DeRoos, and Bayley Tuch.
Most Creative—“Social Distancing Efforts,” Grayson Peters, and Haley Suh.
Most interesting finding—“Paranoia vs. Social Responsibility,” Carter Glenn Gale, Felix Shen, and Danielle Pintacasi .