February 12, 2016
‘Gopher Geeks’ Win Social Data Science Hackathon
A four-person team of Carlson School MSBA students, calling themselves the Gopher Geeks, won the Social Data Science Hackathon held last November at General Mills’ headquarters. The team of Shari Roling, Sharada Vijayalakshman Narayanan, Wenqiuli Zhang, and Priyanka Saboo defeated five other teams to earn first place at the event, sponsored by General Mills, Veritas, Apex, Cloudera, and phData.
The hackathon was organized by Data4Good, a Twin Cities data science meetup group. “The competition involved using data to provide actionable insights to GenerationNEXT on reducing the achievement gap between white students and students of color in the Minneapolis/St. Paul area,” said Roling. “As a former teacher very interested in using data to improve educational outcomes, I knew I wanted to get involved. I asked three classmates to be part of a team, and we became the Gopher Geeks.”
The team was given preliminary sample data to study and determine its strategy ahead of time. On the day of the hackathon, the team used statistical data analysis tools such as SQL, R, and Excel to clean and analyze data using clusters and regression. To further visualize its analysis, the team created heat maps, or graphical representations of data. “Clustering and regression showed several factors that significantly impacted the gap, including proportion of white students to total studenets, expenditures per student, and teacher experience,” Roling said. “We were able to show the size of these factors in schools throughout the metro using heat maps and then overlaid them with a heat map of the achievement gap, showing alignment.”
Before the judges arrived, each team gave a quick overview of its findings to the whole group and received feedback. Then, they had 30 minutes to prepare their final presentation before delivering it to a panel of judges in an analytic Shark Tank-type of competition. In the end, the Gopher Geeks found themselves in first place.
“We learned a ton from this hackathon,” Roling said. “We knew cleaning data was a huge part of working with data, but it really came to life when we spent nearly 70 percent of our time just getting the data in shape to start exploring.” She added that the team worked really well together as each member had expertise in a different area.
“Since I have a background in econometrics, I particularly paid attention to how the industry is using statistical analysis to help obtain insights,” said Zhang. “During the hackathon, I talked to some statisticians and that helped me understand more applications of statistical analysis in industry.”
In winning the competition, Narayanan said that having a lot of preparation gave the Gopher Geeks its edge. “Most of the other teams had interesting analyses, but they had no direction to lead that to actionable insights,” she said. “Since we had done a lot of pre-work before the competition began, we were in a comfortable position to deliver quality insights. And we knew our audience well. We ensured that we portrayed the analysis easy enough for a business audience to comprehend. I think this was another reason that gave us an edge over the other teams.”