Enhancing Student Feedback in Data Science Education: Harnessing the Power of AI-Generated Approaches

Abstract Data science has emerged as an important field in higher education, presenting complex challenges for students due to its interdisciplinary nature involving statistical and computational concepts. In previous research, we introduced the rDSA system (R Data Science Assignments) to support data science students during formative assessments by providing exercises and feedback. As known, formative assessment has evolved from an “interim” evaluation method into a continuous process that enhances student understanding and informs educators to adjust their teaching strategies. It is closely linked to scaffolding, i.e., the method aiming to guide students from their current knowledge to the skills they can develop next. The research reported in this paper builds on previous findings that showed improved learning outcomes and good student engagement. We explored how to enhance student feedback by incorporating Large Language Models (LLMs) with traditional AI into the rDSA system. The results show that students extensively used the system, submitting numerous solutions and attempting various exercises, with activity peaking before exams. The system exceeded expectations in providing useful insights and feedback. However, students faced challenges related to focused attention, probably due to the verbosity of feedback. Learning outcomes, on average, improved, with students who solved more exercises using the tool, performing better than the others.