What Makes an Open Source Code Popular on Git Hub?

The rise of social networks for software development has attached a notion of popularity to open source projects. This work attempts to extract knowledge from the differences between popular and unpopular Python projects on GitHub. A large set of projects was mined for a rich variety of features that measure language utilization, documentation, and code volume. These features were used to train a classifier which predicted current popularity well (F-score = ×8). Notably, these features outperformed measures of author popularity (F-score = ×7). However, these features did not strongly predict future growth in popularity. An in-depth analysis of the perform ant features revealed that they could be useful as a measure of not only popularity, but of code quality.