Through iiD, graduate students mentor undergraduates in labs during the semester and into the summer, bringing them in on intensive, highly creative projects. This hooks undergraduates on the excitement of working with real data early in their Duke education.
In the Data+ program, run jointly by iiD and Duke’s Social Science Research Institute, graduate students lead undergraduates through a deep-dive experience. Each summer, small teams of two or three students work for nine weeks on challenging data sets, often using novel quantitative techniques. Several teams share space, creating a productive, collaborative atmosphere where students learn from and contribute to one another’s work. In 2014, graduate student Chris Tralie and undergraduates Derrick Nowak and Marshall Ratliff used principles and tools from the emerging field of topological data analysis to discover fascinating structural patterns in a large database of musical snippets—from hip hop to classical. Then they built an interactive visualization tool (see www.loopditty.net) to help others understand their findings.
“Casting problems geometrically gives you a lot of information about the data you’re looking at,” says Tralie, a Ph.D. student in electrical and computer engineering.
“One of my undergraduate teams discovered that hip hop is pretty wiggly, and classical and straight jazz tend to be more smooth, but that there is tremendous variation across the board.”
Working in the same room, rising sophomores Alex Pieloch and Carmen Cox used similar techniques to find novel geometric structure in the human cerebrovascular system. This led to publishable work, recently submitted to The Annals of Applied Statistics.