Four graduate students crunch their way to data-modeling title
Graduate | Apr 07, 2016 | Gabelli School of Business
From left, Armi Thassim, Pritha Sinha, Nan (Miya) Wang and John DeMartino – known as The Shooting Stars – are winners of the 2016 March Data Crunch Madness competition.
The four Fordham graduate business students who won this year’s March Data Crunch Madness competition called themselves “The Shooting Stars.”
And just like the Villanova University basketball player who poured in a three-pointer with no time on the clock to win the NCAA championship this year, the stars at Fordham shot with deadly accuracy when it came to data modeling.
The competition featured more than a dozen teams seeking to create a data model that would predict the results of the NCAA men’s basketball tournament. What distinguishes the success of The Shooting Stars – Armi Thassim, Pritha Sinha, Nan (Miya) Wang, all MSBA ’16, and John DeMartino, MSIS ’16 – is that 75 percent of the team knew nothing at all about basketball.
“When we first sat down, I had to explain what basketball was: the rules, the concept of the tournament,” said DeMartino, the lone team member who knew the sport.
That didn’t hinder the team, all of whom knew one another from their data-mining class.
Using well-developed machine-learning techniques and data-mining skills, the team modeled with high accuracy: 78 percent, compared with 73 percent for last year’s winner.
Among the techniques used to improve their model were feature engineering, algorithm optimization, cross validation, and model ensemble, Wang explained.
“We’ve used Sklearn, a machine learning python package to fulfill the tasks,” Wang said. “About 2,000 lines of codes were written to do the work: data cleaning, model building and output visualization.”
The winning team was awarded $750, the second-place team won $500, and the third-place team took home $250. A $50 award also was given to each of the other teams that participated.
The competition was spread out over a number of weeks, with historical data on the tournament being released to students in February and this year’s team data given out in March.
Gabelli School of Business Dean Donna Rapaccioli said the competition was a great exercise in teamwork for the students.
“When you go out into the workforce … you’re going to realize that teamwork is really key, and this experience hopefully allows you to get to the next level in being a good team member,” she said.
