Mine Çetinkaya-Rundel (University of Edinburgh, Duke University, RStudio) & Tiffany Timbers (UBC)
As enrolments in statistics and data science courses grow and as these courses become more and more computational, we as educators are faced with an interesting challenge -- providing timely and meaningful feedback. The simplest solution is writing assessments and assignments that are easier to auto-grade, e.g. multiple-choice questions or coding exercises with a single correct answer. While these types of exercises can be valuable, we can't envision assessing mastery of the entire data science cycle using only these types of exercises. In this session, we will present our experience using learnr tutorials, nbgrader, and continuous integration tools like GitHub Actions to provide immediate feedback to data science students. We will provide examples of exercises and feedback, discuss design choices as well as opportunities and challenges presented by working with these systems to supplement human feedback in data science courses.
Slides for the talk can be found here and they contain links to demos we will walk through during the talk. Source files for all materials presented and referred to during the talk can be found in this GitHub repo.
