A Master's with a Difference
Over the past five years, scores of universities around the country have introduced master’s-level data science programs, but few universities approach data science the way the University of Virginia does. “Most data science programs are housed in a specific school or department,” notes Donald Brown, director of the University’s Data Science Institute and the W. S. Calcott Professor of Systems and Information Engineering. “If they’re in a business school, students get a business analytics take on the subject. If they’re in a computer science department, students focus on machine learning or programming.”
U.Va. does data science differently, providing a broad interdisciplinary perspective. “The Data Science Institute engages all aspects of the University, but does not fall within any particular school,” observes Arlyn Burgess, the program’s executive coordinator. The Master of Science in Data Science degree is conferred by the Data Science Institute, which reports directly to the Provost’s Office.
Faculty from three separate departments — Computer Science, Statistics, and Systems and Information Engineering — teach core courses in the accelerated, 11-month curriculum. The curriculum also includes the choice of a broad range of elective courses, a significant capstone project, and a course on the Ethics of Big Data, something that is rarely offered in similar programs.
“We’re able to look at the way data science is practiced in a wide variety of fields and bring ideas and techniques from these fields to the classroom,” Brown says. “Students leave the program knowing a number of ways to address a particular data science challenge.”
This interdisciplinary approach has in turn attracted students from an unusually wide range of backgrounds. “The first cohort, which graduated in May, included students from the less data-intensive fields like history, political science, biology and economics,” says Jeffrey Holt, a professor of statistics and the program’s director. “They bring a variety of interesting perspectives to the classroom.”
These perspectives are reflected in exceptional variety of capstone projects that the first cohort worked on throughout the year. They included techniques for early sepsis identification in hospital patients, a new, more accurate model for predicting energy consumption in U.Va. buildings and the development of a game that emergency responders can use to simulate entering and maneuvering in a building during a crisis. The teams working on sepsis detection and energy prediction won best paper awards at the Engineering School’s annual IEEE symposium in April. “We actively solicit projects from industry and government to ensure their real-world relevancy,” Brown says. “For instance, Leidos, a defense contractor, provided funding for the smart metering project.”
The program’s distinctive focus has attracted attention. “We had stronger than expected applications for the inaugural cohort and stronger-than-expected acceptances,” says Kevin Skadron, the chair of the Department of Computer Science and, along with Holt and Brown, one of the chief planners of the program. “The response only improved this year, suggesting that the program is meeting a need.”
On the basis of feedback from first-cohort students and industry advisors, the Data Science Institute has already made a number of small adjustments to the curriculum. It made the capstone a full-year experience and is increasing its emphasis on the use of SQL for database queries. It has just hired a career services specialist, affiliated with the University’s Office of Career Services, dedicated to helping place the program’s graduates.
“With a program like this in such a fast-evolving field, continual change is the order of the day,” Brown says. “We’re always thinking about ways to modify the program to ensure it remains as current and as valuable to students as possible.”