Image: The first spring football practice of the season, March 20, 2025. Photo by Duke Athletics.


Project introduction

Duke Athletics has provided physiometric data, measured longitudinally, for a number of football players using four instruments from a recent year. Your objective is to use these data to come up with interesting data insights that might not be immediately obvious from surface-level analyses (note that you do not need to use data from all instruments).

For instance, is relatively poor performance on one apparatus associated with poor performance on another apparatus? Are there multivariate patterns in physiological measurements across instruments? Do you detect correlations with performance through time (perhaps after game days, etc., by incorporating external data)? Can you perform a clustering-based analysis in which players might be grouped in interpretable, but not necessarily obvious ways?

Note that there is no “target” variable to perform prediction or inference on - the purpose of the project is to explore relationships in a wide variety of real-world data sources.

This is a challenging, open-ended problem using a type of data you might not have seen before. The data, despite some cleaning, will be messy. Care should also be made to use readily-interpretable techniques, with conclusions and interpretations able to be understood by the data providers.

In general, guidance for a successful project is to answer the following question: what type of conclusions can you draw from the data that might lead to actionable insights for Duke Athletics?

Learning objectives

  • Solidify skills in reproducible research and programming, including version-control and collaboration via GitHub
  • Critically think about reasonable analysis approaches in the context of real-world data
  • Describe statistical models and/or approaches clearly and correctly
  • Develop scientific writing skills by providing clear, concise, data-driven conclusions suitable for allied researchers

Data matters

Nick Potter, Director of High Performance & Sports Science for Duke Atheletics, presented the data on Tuesday, March 18. He was assisted by Anmol Sapru, who is more familiar with the minutiae of the data themselves. There is an ongoing open assignment in Gradescope for you to submit questions, which will be given approximately weekly to Anmol - answers will be provided with an FAQ here. Finally, the user guides for three of the four apparatuses are included in your GitHub repositories - you may gain further understanding for the variables through those user guides.

For grading details and other logistics, refer to the README in your repositories.

Project timeline

  • Group: Report and reproducible code
    • Due Friday, April 4
  • Group: Revised report and response to reviewers
    • Due Tuesday, April 22

Note: each team’s GitHub report repository and commit history will also be evaluated by the instructor. The GitHub repository must contain the reproducible R Markdown document corresponding to the submitted reports, and will be checked throughout the course of the case study to ensure all team members are making meaningful contributions to the project.