Mining Undergraduate Students’ Code Repositories: Insights from Interdisciplinary Software Projects

Abstract

Due to its multidisciplinary and dynamic nature, it is challenging to design Software Engineering (SE) educational material. To do so, universities must consider the complex working environments; recent technologies; and tools and skills, in order to prepare students to fulfill the expectations of the software industry. This study was carried out in a Brazilian private university, specifically in courses called Interdisciplinary Software Project (ISP) of the SE major. These courses are project-based, conducted by two professors at the same time in classroom, following a Scrum-like process. The objective of this study was to characterize how students work and collaborate in a group environment where agile development is used as well as how their behavior reflect on the use of Software Configuration Management (SCM) practices. In order to achieve this objective, the study analyzed 38 students’ code repositories. This study has found out that students procrastinate sprint work, since 51% of the commits are performed when less than 20% of the sprint time is left. We have also observed that in 87% of the groups just one member is a top contributor and would harm the projects’ outcomes if he/she left the project. In terms of SCM practices, we have identified that most commits comprised changes in up to three files. Moreover, most commit messages are less than 10 words long. This could mean that students do not commit many alterations at the same time but could make better use of messages in order to facilitate the comprehension of alterations by other members.