Abstract

Effectively staffing a software development team for a software project is important to the development of software in the project. Performance of software developers can be measured by the quality of the produced software, and the number of code smells is one factor that indicates software quality. However, modern software development uses distributed version control systems in which different software developers collaborate to develop software. The number of code smells in the software is hence the result of the aggregate performance of the whole team. This makes it difficult to measure the performance of individual developers. This paper proposes a method and a supporting tool for measuring the performance of individual software developers in a Git project based on code smells. Bayesian average rating is adopted to rate the performance of each developer in a project by taking into account his/her level of contribution to the project, i.e. the number of source code commits, as well as his/her effort to produce clean code. An experiment on C# projects shows that there is a strong positive correlation between ranking of developer performance in a project by the proposed method and that by human evaluators.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call