A team of four Cornell College students received third place in an analytics competition that involved teams from across the country.
Clara Haverstic, Asher Muse, Chase Sonnemaker and Sabrina Honigman participated in the MinneMUDAC Student Data Science Challenge and submitted their work on March 19.
According to the website, student teams built predictive models to predict the outcomes of the 2021 NCAA Division I Men’s Basketball Tournament. Teams constructed their tournament bracket using their predictive model.
“Our students’ success is the result of persistent hard work in and outside the statistics classroom, setting a wonderful example for all Cornell College students,” said Cornell assistant professor of statistics Tyler George.
“I think we succeeded because we took a step back and thought about the problem as a whole,” Mr. Muse said. “We were patient and took our time understanding the problem before jumping in, which made the analysis and model creation a very smooth process.”
The team started by researching what other participants who won the tournament had done in the past.
“Their techniques were strong but ultimately difficult for us to implement as they used highly advanced algorithms,” Mr. Sonnemaker said. “We decided to start with some of the other techniques we were familiar with, ultimately using Logistic Regression to create a model predicting the probability of a team beating another.”
The team applied many skills from their Cornell classes over the years.
“We worked with the R programming language, which we all learned a lot about in the Introduction to Data Science course,” Ms. Honigman said. “We also used many of the modeling techniques learned in Statistics 201 (Statistical Methods I), like logistic regression.”
Ms. Haverstic, a sophomore with an individualized major in political analytics, said she learned a lot from the seniors on the team. Mr. Sonnemaker, Mr. Muse, and Ms. Honigman just graduated and said they’ve participated in many MinneMUDAC challenges, but this was their best finish.