Automated Feedback to Students in Data Science Assignments: Improved Implementation and Results

Abstract

The automated grading of assignments is a long discussed topic in the field of technology-enhanced learning. In such a large research area, the authors focused on the automated grading of assignments made up of a mix of commands (in R language), their output and comments (in natural language). In particular, the paper discusses several improvements on the automated feedback generated by a tool developed at the University of L’Aquila, to support the students during their study of the subject. The goals of the research are the implementation of a feedback that gives an explanation of the automated grading, also providing students with the causes of the mistakes and suggestions on how to correct them. Accordingly, we designed and developed an automated feedback, used by students during the current academic year to support their homework. We then collected the students’ opinions through both standardised and ad-hoc questionnaires, so to evaluate the effectiveness of our proposal and identify the aspects to improve. The results highlight an increased engagement while performing the assessment, the usefulness of the feedback, as well as where the explanation was clear and where improvements are needed.