As DataFest hosts, Maria Tackett and Alexander Fisher from Duke University share their reflections on how the event challenges students to think critically and work collaboratively under pressure.
Real data is big. Real data is messy. The students in today’s classrooms will go on to work with real data daily in their professional and personal lives. Despite this, it is challenging to provide many opportunities for students to work on open-ended applied problems as complex as the ones they will experience outside the classroom. While internships and research assistantships can help fill this void, the stakes are high and the opportunities are few. At Duke DataFest, we strive to engage as many students as possible, from multiple institutions and a wide variety of backgrounds, in real data analysis in a low-stakes environment.
Our task begins with outreach several months before the event, as we send email announcements to students in statistics, computer science, and other STEM programs. We make in-person announcements in the introductory statistics courses to encourage students early in the curriculum to take part. We also send email announcements to a variety of affinity groups across campus to expand our reach to students who may be interested in data-driven work but not currently enrolled in statistics and related courses.
Additionally, we recruit faculty at other local institutions who are interested in being faculty sponsors for Duke DataFest. The faculty sponsors not only play an integral role in recruiting students at their home institutions but also volunteer time at the DataFest event. Our hope is that by being physically present at DataFest, the faculty sponsor evokes an esprit de corps among the students who do not attend Duke.
Our main partner institution, The University of North Carolina at Chapel Hill, has also reimbursed local student travel to our campus to encourage participation.
Fundamentally, DataFest is a team event; however, not all students who wish to take part sign up with a team. We give students the option to sign up individually and be assigned a group. We sort students in groups based on a survey that asks students about their anticipated engagement with the event and their coding language of choice. We expect these two criteria will help reduce team conflict and promote collaborative computing. In fact, some of these groups go on to win awards at DataFest.
For students who would like to better prepare for the event, we offer optional workshops on topics such as data visualization, regular expressions, and Tableau in partnership with Duke library staff. Additionally, pre-recorded workshop sessions are hosted on the Duke DataFest website to help students new to data science learn the basics of wrangling data in R and Python.
Once DataFest begins, our goal is to create an environment in which students can focus on the data task, persevere, and present their findings at the conclusion of the event. We do this, in part, by making it possible for students to be physically present the majority of DataFest weekend. We provide free meals to participants, as well as an abundance of snacks, throughout the event. Moreover, we have a variety of activities between meals—group photos, trivia, guided stretching exercises, and small prize giveaways—to help increase participation.
To streamline onboarding into the competition, we partner with NCShare and the Duke Office of Information Technology to provide all participants containerized versions of RStudio and Python notebooks that can be accessed through a web browser. These containerized environments come pre-loaded with the DataFest data set, documentation, and a suite of data wrangling packages. This enables students to avoid debugging local installation of software and simply focus on DataFest. Students who opt to use their local computing environment download the data set and documentation from a private Box link.
In the past three years, we’ve had 180 teams take part, representing six schools in the area. In 2025, we had 57 teams made up of 161 attendees participate in Duke DataFest. Of these 57 teams, 31 submitted a final presentation.
Read more perspectives from DataFest hosts:
This brought back great memories! Working through massive, messy datasets overnight can be intimidating, but having workshops, mentor consultations, and pre-loaded cloud environments makes the learning curve so rewarding.
see