Corrective commit probability: a measure of the effort invested in bug fixing

Idan Amit*, Dror G. Feitelson

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

5 Scopus citations


The effort invested in software development should ideally be devoted to the implementation of new features. But some of the effort is invariably also invested in corrective maintenance, that is in fixing bugs. Not much is known about what fraction of software development work is devoted to bug fixing, and what factors affect this fraction. We suggest the Corrective Commit Probability (CCP), which measures the probability that a commit reflects corrective maintenance, as an estimate of the relative effort invested in fixing bugs. We identify corrective commits by applying a linguistic model to the commit messages, achieving an accuracy of 93%, higher than any previously reported model. We compute the CCP of all large active GitHub projects (7,557 projects with 200+ commits in 2019). This leads to the creation of an investment scale, suggesting that the bottom 10% of projects spend less than 6% of their total effort on bug fixing, while the top 10% of projects spend at least 39% of their effort on bug fixing — more than 6 times more. Being a process metric, CCP is conditionally independent of source code metrics, enabling their evaluation and investigation. Analysis of project attributes shows that lower CCP (that is, lower relative investment in bug fixing) is associated with smaller files, lower coupling, use of languages like JavaScript and C# as opposed to PHP and C++, fewer code smells, lower project age, better perceived quality, fewer developers, lower developer churn, better onboarding, and better productivity.

Original languageAmerican English
Pages (from-to)817-861
Number of pages45
JournalSoftware Quality Journal
Issue number4
StatePublished - Dec 2021

Bibliographical note

Publisher Copyright:
© 2021, The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.


  • Corrective commits
  • Corrective maintenance
  • Effort estimate
  • Process metric


Dive into the research topics of 'Corrective commit probability: a measure of the effort invested in bug fixing'. Together they form a unique fingerprint.

Cite this