Logo image
Lost Gems: Predicting Module Abandonment on Community Hosted Open-Source Repositories
Thesis

Lost Gems: Predicting Module Abandonment on Community Hosted Open-Source Repositories

Chen, Li Chan
Masters, 國立清華大學, 服務科學研究所
2015

Abstract

模組 停止維護 GitHub RubyGems 資料探勘 決策樹 邏輯式迴歸 隨機森林 Module Abanodnment RubyGems GitHub Data Mining Decision Tree Logistic Regression Random Forest
Open source modules, which are software libraries, are often abandoned by their maintainers. This abaondonment becomes a serious concern for developers and for software projects at large. In this study, we want to first help developers recognize module abandonment in a data-driven manner. Second, we wish to help the sites which host modules to examine the value of preserving expensive usage data. We examine the modules of the Ruby programming language in this study. We try to predict if gems will be abandoned in the near future by analyzing data from GitHub and RubyGems using data mining techniques. We discover effective models to predict abandonment and factors associated with it. Moreover, we begin to percive that it may not be necessary to collect expensive usage data from independent module hosting communities to predict module abandonment. Using these results, we build an online dashboard to help Ruby developers recognize potential abandonment of modules. This study helps us to combined theoretical methods with practical use.

Metrics

1 Record Views

Details

Logo image