rDSA : an intelligent tool for data science assignments

Tools supporting the teaching and learning of programming may help professors in correcting assignments and students in receiving immediate feedback, thus improving the solution before the final submission. This paper describes the rDSA tool, which was designed, developed, and evaluated to support students in completing assignments concerning (i) the execution of statistical analyses in the R language and (ii) commenting on the results in natural language. The paper focuses on the feedback provided by the tool to students and how it was designed/evaluated/improved over the years. The paper also presents the results of two studies that indicate the advantages of using the tool in terms of engagement and learning outcomes. To conclude, we provide a discussion on the characteristics of the tool, its use in similar courses, and the scope for future work.