This paper analyzes the evolution of open-source software (OSS) from two perspectives—“commit history” and “developer behavior”—using the statistical-analysis OSS textit{PyMC} as a representative case, with the aim of deepening our understanding of OSS evolution based on these findings.
To achieve this purpose, we pose three research questions:
(1) To what statistical model does an OSS commit history conform?
(2) How do developers concretely contribute to an OSS project that changes daily?
(3) Can an OSS development process be simulated on the basis of the developer behavior identified in (2)?
We have developed metrics that quantify both the workload and the work content of each committer, and have clarified concrete developer contributions to PyMC from the commit history.
In future work, we plan to run computer simulations of OSS development based on the developer behavior thus identified.