Final Thesis: Predictors of Successful Open Source Projects
Abstract: The number of open source software projects has continued to grow for many years now, with a portion of them being very successful and broadly adopted. For companies and individuals involved in an open source project, it is advantageous to estimate its success and longevity. There are some high-level models of success to be found in literature. However, other than growth, factors of success derived from commit data have been barely studied by academics. This thesis searches for predictors that can contribute to a predictive model. We use an exploratory approach on a large set of over 11,000 active projects to find predictors. During the course of this work, we implemented a duplicate correlation filter in order to account for forked projects, and we designed a general ranking metric. We compared the distribution of features between the high ranking projects and all others. Through that, we found that the amount of initial contributors and the number of active months correlates positively with success. An in-depth evaluation of the explored predictors is not part of this work.
Reference: Jonathan Frieß. Predictors of Successful Open Source Projects. Master Thesis, Friedrich-Alexander-Universität Erlangen-Nürnberg: 2019.