Students competed in teams to present the most informative and sophisticated data analysis at DataFest, the University’s first data analysis competition that was held March 28-30.
The competition had 46 participants, and although 13 teams had signed up for the event, only seven participated.
Duke and UCLA have had similar competitions in the past and for this competition, and Duke provided the common data set that all the teams were required to analyze.
The competition was sponsored by the Operations Research & Financial Engineering department, along with the Princeton Statistics Laboratory, the Computational Stochastic Optimization and Learning Department of Operations Research, RJMetrics, and the IBM Academic Initiative. It was organized by assistant professor Han Liu, professor Deborah Nolan and professor Philippe Rigollet of the ORFE department and was held in Sherrerd Hall.
Rigollet noted that this year’s problem and data set — information regarding energy usage from over 100 buildings — was provided by Gridpoint, an energy management company that places smart meters in medium to large businesses. Teams took different approaches to using the data, mostly by studying the effects of energy consumption.
On Sunday each team was given four minutes to present their analysis of the data to a panel of judges from Google, PSE&G, RJMetrics, IBM and the University faculty. Each team was allowed three slides to provide visual aid for their presentations and submitted a one-page report of their work. A $50 cash prize was awarded to the winning teams in each category.
Throughout the duration of the competition, one-hour technical sessions were given to the competitors on topics such as nonlinear modeling and machine learning to aid the competitors in their analyses of the data set. Teams were also able to consult with mentors.
“The participants were really fun to be around,” Rigollet said. “Some students put in 50 hours of work, but it was still pretty laid back.”
Liu, the primary organizer of the event, said he was pleased with the turnout of the competition and the diversity of the competitors. Fewer than 20 students were from the ORFE department, Rigollet noted.
“A lot of students attend from different departments of the University, and everyone was happy,” he said. “I think this is a great event for the University to provide the students the opportunities to touch, actually touch, and explore, and analyze very complex data sets. I saw a lot of enthusiasm, so this is a very successful event.”
Student participants also remarked that the competition was a rewarding experience.
“I think it was a great learning experience. My team consisted entirely of freshmen, so we had to pick up new skills and analytical methods on the spot,” Lydia Liu ’17, a member of one of the winning teams, said. “The best part of the competition was that we had a lot of help from the graduate students, the professors and industry experts. I’ll definitely consider doing it again next year.”
Rigollet said that the department plans to hold the event again next year.