Students Hack Real-World Data at DataFest

May 11, 2015

Hundreds of statistics undergraduates; one giant, real-world data set; and 48 hours to come up with something good — welcome to DataFest. Two teams of Cal Poly statistics students traveled to UCLA in April for this statisticians' version of a hackathon, competing against 50 teams from five universities. Though top honors went elsewhere, students enjoyed a unique opportunity to work with actual data supplied by industry sponsor edmunds.com, a service that helps customers shop for cars.

"It's a really good chance for the students to get their hands on a much larger data set than we usually get to offer them," said Hunter Glanz, faculty sponsor for the two teams. "This was around five million observations, so it's very real and challenging."

The competition invites teams to get creative with the demographic, web traffic and sales data provided. Students were given the broad question, "How can we make car shopping easier?" and then given free rein to figure out their own approach to the data.

"Translating this overarching idea or business question into statistics questions about the data is a really valuable skill," Glanz said.

"DataFest really embodies Cal Poly's Learn by Doing philosophy," said Phillip Richardson, one of the team members. "There is no step by step guide for how to analyze this data. It requires poking, prodding and getting your hands dirty. Literally the only way to learn from this data is to dive in."

Teams competed in one of three categories: best recommendation to the company, best visualization and best use of outside data. Cal Poly competed in the best recommendation and best visualization categories. The teams' conclusions are subject to a non-disclosure agreement with edmunds.com.

"It's like trying to solve a puzzle but without having a picture to go on and not having all the pieces. You have to use the tools you've acquired during your time at Cal Poly to build the pieces and come up with the puzzle picture yourself," Richardson said.

The event also provided a networking opportunity for students with a number of employees from edmunds.com both attending and acting as judges.