Data Mining Tool to Discover DevOps Trends from Public Repositories: Predicting Release Candidates with Gthbmining.Rc

• Daniel D. R. Barros UFABC
• Flávio Horita UFABC
• Denis G. Fantinato UFABC

Resumo

Public repositories have been performing an essential role in bringing software and services to technical communities and general users. Most of the cases, public repositories have a DevOps tool, with a live and historical database behind it, to support delivering and all steps this software or service should adopt before going to production. This paper introduces gthbmining, a data mining set of tools to discover DevOps trends from public repositories, and presents the module gthbmining.rc. Considering the premise of a GitHub public repository, the main contribution here is predicting release candidates, an important label a software release has. The methodology, architecture, components and interfaces are explained, as well as potential users. The results show a reliable and flexible tool, as classifiers metrics and graphics are provided, along with the possibility to add new data mining algorithms in the open source module presented. Related works are also supplied, and a conclusion shows the outcomes gthbmining.rc can provide.

Palavras-chave: Release Candidate, GitHub Mining Tool, DevOps, Data Mining