Creating Bucknell’s final exam schedule is an intricate challenge, requiring the registrar to organize exams across 80 time slots for about 3,200 students. For 15 years, Vince Pellegrini, assistant registrar, has tackled the complex problem, working to develop an exam schedule that minimizes “inconveniences” for students and faculty — such as back-to-back exams or three exams scheduled in 24 hours. With his analog approach, it took up to three months to produce each schedule. Fortunately, at Bucknell, such problems are learning opportunities for students.

A team from Bucknell’s three colleges partnered with Pellegrini, University Registrar Tim Kracker and Data Analytics Architect Mike Latorre to devise a more efficient method that also reduces inconveniences for students.

Pellegrini took his challenge to Sam Gutekunst, the John D. and Catherine T. MacArthur Professor of Data Science, who reached out to Lucas Waddell, mathematics, and Thiago Serra, analytics & operations management, to collaborate. Each invited a student to join the team. “The number of possible exam schedules is massive — far more than there are atoms in the known universe,” says Gutekunst. “Bringing students into a project of this scope that also had a direct application to the University is what made it so exciting.”

In 2023, Stanley Gai ’25, business analytics and sociology, created a dashboard illustrating enrollment data, which Pellegrini used to refine the fall semester schedule. Clara Chaplin ’25, mathematics and computer science, and Tsugunobu Miyake ’25, computer science & engineering, developed a mathematical model and programmed a user interface that generates schedules with minimal inconveniences.

To develop the spring 2024 schedule, Pellegrini used the computer-generated sample schedules as his starting point, which accelerated the process; he estimates it saved him several weeks of work on exam scheduling. He then applied his insights about campus dynamics and factors not easily quantifiable by data alone. “Vince is the one who ultimately knows the nuances we can’t program or predict,” Miyake says. Inconveniences will never be entirely eliminated; human expertise will ensure the schedule runs smoothly.

The program drastically reduces the time Pellegrini dedicates to exam scheduling. Comparing his traditional method with the new collaborative approach shows striking results. In the spring 2023 schedule, 1,182 students experienced at least one inconvenience. In the schedule for spring 2024 — generated using Gai, Chaplin and Miyake’s work — that number was reduced to 432. “My job will never be the same,” Pellegrini says. “I’m amazed at what we have at our fingertips.”