Final Thesis: Classification of Commit Characteristics by Code Changes
Abstract: Understanding the types of changes in Git commits plays a crucial role in calculating development metrics and improving organisational decision-making. However, even though the usefulness of classifying commits is widely recognised, there has been no consistent approach to doing so, especially not at classifying commits in hindsight. In this thesis –using a design science approach– we want to present a new approach that can be used for automatically classifying commit changes. Furthermore, we developed a widely applicable set of classifications based on previous attempts to classify different types of commits. Both the classifications and the model are constructed using openly available literature and GitHub repositories. In addition, we demonstrate the functionality of our approach by classifying a set of commits.
Keywords: Code classification, machine learning, metrics
PDF: Master Thesis
Reference: Philipp Kramer. Classification of Commit Characteristics by Code Changes. Master Thesis. Friedrich-Alexander-Universität Erlangen-Nürnberg: 2024.