Industry Day—a celebration of the first Data Science Competition, organized by the statistics department along with data scientists—was held virtually on April 28. Following introductions by Provost Jack Hu and Alan Dorsey, dean of Franklin College of Arts and Sciences, the top undergraduate and graduate data science teams presented their findings from the competition.
Industry Day included a panel discussion about careers in data science from brand executives before awarding the winning undergraduate team, led by Bryce Davis, and the winning graduate team, led by Jeevan Jankar, for their work.
“We can directly explain to the customer where they go wrong, what you can do better and how your past behavior influences current, automated decisions,” Davis said. “That’s how we won the competition.”
Both runner-up and winning teams at the undergraduate and graduate levels gave brief presentations on their project research processes during the competition.
Each team was tasked with developing a quantitative model for making credit card decisions. They received simulated datasets typical of performance of credit card accounts with credit history information.
Using that information, every team developed two models, one of which used traditional statistical methods and the other used a modern machine learning algorithm. The individual teams compared the results from both approaches, chose an appropriate method and used it to create an algorithm for making accept and decline decisions on credit card applications.
Two of the runner-up undergraduate teams—one led by Orr Shalev and one led by Ayush Kumar—discussed their methods and experience from the competition.
“Funnily enough, we learned so much about how much we still have to learn,” Shalev said. “We were well aware that there are almost an infinite amount of ways to pick apart the task that was presented to us, and given the limited time frame—just like in real-life scenarios—we need to do our personal best effort based on what we know.”
The winning graduate team and the runner-up graduate team, led by Zhizhong Lin, spoke about their respective credit card approval models.
“We were definitely very glad and felt like our hard work paid off,” Will Cranford, a member of the winning graduate team, said. “The three of us worked together on a group project over the summer, so we knew the group had good chemistry, and it worked well this time around.”
Vijay Nair, one of the industry partners from Wells Fargo, said he was beyond impressed with the work he saw from the students.
“We wondered if this would be too hard, particularly for undergraduate students,” he said. “I must say, I’m floored—totally, totally floored by what a great job you guys did, OK?”
T.N. Sriram agreed and expressed the reasons for hosting Industry Day.
“This was the point of holding this Industry Day: We wanted to showcase the talent we have at UGA,” he said.
Along with Nair, other industry partners gave advice to the students. Khalifeh Al Jadda, the director of online core data science from Home Depot, spoke as a panelist.
He shared advice about succeeding in the industry as a successful data scientist.
“You absolutely deal with small datasets in the classroom, so the biggest difference you’re going to face when you go into the industry is the scale of the data that you deal with and the scale of the production,” he said. “Always when you build any model, you’ve got to think, ‘This model is going to now serve millions of customers.’”
Susan Daniel, a senior microbiologist from Johnson & Johnson, reminisced on what the data science industry looked like decades ago.
“I was drawn to data science because our analytical capacities have increased significantly over the last 20 years. Now we’re able to understand giant datasets affordably and efficiently with the use of data science,” Daniel said. “Back in the ’90s, that wasn’t so much the case. You could pull your data up in Excel, and it wasn’t a problem.”
Amil Williamson, who currently works as a data scientist at Georgia-Pacific, graduated from the University of Tennessee in 2019 and was asked to speak on the transition from graduate school to the industry.
“To be honest, the transition has been relatively seamless, but I attribute that to joining a great team. Everyone’s purviews are in kind of different areas, and that’s the beauty of working on a team,” Williamson said.
Having an academic basis in addition to practical experience is very valuable, Daniel said. In the same vein, she also spoke on the benefits of graduate school.
“The industry is starting to be able to discern between the technician data scientists … and the expert data scientists,” Daniel said. “Having a deep understanding of data manipulation and statistics is incredibly valuable to the end consumer, whether it be a patient or a commerce-retail consumer. Having a master’s degree or advanced coursework is a great way to demonstrate your competency.”
Mohamad Al Lawati, who sponsored the event, gave closing remarks and reiterated how impressive the resulting projects were.
“I want to thank the UGA students for participating,” he said. “All of them deserve a lot of credit for their dedication and hard work. I want to extend my warmest congratulations to … the winners of the competition.