Predicting long-time contributors for GitHub projects using machine learning
Open AccessOrganizations typically develop software systems using non-developmental items and commercial off-the-shelf software. Many organizations are changing the way they create, capture, consume, and commercialize software by increasingly Open-Source Software projects. However, many OSS projects do not survive, as the survival of OSS projects depends mainly on retaining new contributors. While organizations driven by their business needs support some OSS projects, most OSS contributors are volunteers who contribute their work for free. While many join OSS projects, only a few contribute to an OSS project for a considerably long time. A Long-time contributor is defined as a contributor who joins a project and continues to contribute for more than T years; T is generally set to 1, 2, and 3 years. LTCs often contribute more code than non-LTCs.Most new contributors abandon a project without becoming LTCs. The data in this research shows that 98% of new contributors leave a project before three years. Various factors affect whether a new contributor becomes an LTC, including the new contributor's experience, expertise, the project's maturity, working environment, documentation, and task difficulty. Identifying factors that predict potential LTCs can enable project owners to gain insight into what matters to contributors and take action to retain new contributors for a long time. These actions could include mentoring, quickly responding to questions, providing timely code reviews, and merging contributions.This research investigates effective predictability of new contributors to OSS repositories becoming LTC based on repository and contributor meta-data collected from OSS repositories. Compared to state-of-the-art models, the models built in this research use less than 50% features and produce better results. In 10-fold cross-validation, the precision, recall, F1-score, MCC, and AUC of the best state-of-the-art model are 0.546, 0.041, 0.075, 0.1446, and 0.908, respectively.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.