Georgia Statistics Day shines spotlight on student research, industry connections

Georgia Statistics Day is an annual event, designed to promote interdisciplinary research among the three premier academic institutions in the state of Georgia – UGA, Emory, and Georgia Tech – initiated by statistics units at the three institutions in 2015 to emphasize mentoring of junior researchers and interaction between industry and academia.

The 2025 iteration delivered that and more. The one-day conference enabled junior researchers to present their work, to see state of the art developments in research on statistics and related scientific areas, and to interact with some of the key players in the sector. “The informal setting encourages the free exchange of ideas and stimulates research in topics of practical relevance to science and industry," said Dan Hall, professor and head of the UGA department of statistics. 

The primary activities included technical sessions featuring academic research in statistics and data science by faculty at the three rotating host institutions/departments. Chief among these was the Keynote Address, which was delivered by Liza Levina, the Vijay Nair Collegiate Professor of Statistics at the University of Michigan. 

The Student Poster Session in the Pecan Tree Galleria featured the work of more than 52 students, including undergrads and graduate students from the three main institutions, but also from Clemson and Georgia State University. Awards were given to eight submissions, including three undergraduates and five graduate students. Among the UGA honorees were statistics major Deeya Datta, data science major Sloka Sudhin, and Ph.D. student Jiazhang Cai. 

An Industry Session featured talks from sponsors JMP and Unclaimed Property Consulting and Reporting (UPCR) focused on careers and job/internship opportunities at those companies.

The variety of research inquiry on display during the poster session speaks to the ambitions of statistics students across the region as well as the central role of data science across disciplines and sectors.

Jingying Gao, a UGA doctoral student in statistics, shared her research poster on tail index estimation and its uses to classify stocks as low, medium, and high risk.

The phenomena of big data supplied the context for many of the research endeavors. UGA statistics Ph.D. student Yumo Peng explained the need for supervised subdata selection in terms of the failures of existing models in the presence of imbalanced data classification, with applications ranging from detecting fraud in finance to diagnosing rare diseases. “I think we’re dealing with an overflow of data, so you’re always trying to figure out what’s the value of the information,” said Peng, who completed her undergraduate degree from UGA, attended Ohio State for her MS, and returned to UGA for her doctorate. “In any case, you have to be careful about what data you’re inputting into your model.”

Many students presented work that is a product of years of research and training, even as undergraduates.

“It's a software package I’ve been working on for around two years, and it uses particle swarm optimization to find the optimal designs for mixed-factor experiments under generalized linear models,” said Sloka Sudhin, a fourth-year data science and statistics major from Marietta. The research grew out of a textiles experiment by her research mentor Joshua Lukemire, Assistant Research Professor at Emory University Rollins School of Public Health, while he was an undergraduate at UGA. 

“They had some factors that were continuous, some that were discrete, and they were measuring a non-normal response, which caused the issue,” Sudhin said. “So, we set out to create a software package for designing experiments that’s able to find these optimal designs, for any kind of experiment. We determine what’s optimal using a criterion called D-optimality – it uses the information matrix of a design, and if you’re able to minimize the determinant of the matrix, then you can think of it as minimizing the variance of your experiment, before you even conduct it. We use numerical approximation models and turn that into an objective function and use particle swarm optimization, a technique similar to the way birds scan a field for food.” 

A Double Dawg in statistics, Foundation Fellow and 2025 Goldwater Scholar, Sudhin’s interest in statistics coalesced in Abhuyuday Mandal’s FYOS course. She entered the honors program and continues to work with Mandal on design theory.

“I have very high expectations, and I give them the big picture of what they can accomplish,” said Mandal, UGA professor of statistics. “I want them to dream big, and I let them know they can do great things, and I ask them to believe in themselves as they discover new opportunities. This event is a wonderful example of that hard work on display.” 

Image: Participants at Stat day 2025, from left: Joshua Lukemire, Sloka Sudhin, Abhyuday Mandal, senior data science major Audrey Floyd, and Deeya Datta. Photo by Jason Thrasher.