Analyzing the Growth Patterns of GitHub Projects to Construct Best Practices for Project Managements

The Git platforms, such as GitHub, are big data providers that store various development histories. Then, many researchers analyze data from such platforms from various aspects. Recently, AI-based systems have been used to construct themselves. However, there are no studies to measure the quality of the projects, and the ideal images of the projects need to be defined. This paper aims to find the ideal images of the projects in OSS (Open Source Software). For this, we extract the time-series project metrics from the famous OSS projects to categorize for detecting the patterns of projects’ growth. Our approach tries to explain the patterns and to give the decision for the patterns, good or not. The time-series metrics from projects include the number of stargazers, forks, commits, etc. The number of stargazers should increase as time passes, and the number of forks tends to decrease. The stargazer pattern indicates that many developers watch the repository since it was managed well. We conducted the case study to analyze the time series data from 10 repositories in GitHub. As a result, we found that the transitions of the number of issues typically form the sawtooth wave. The sawtooth wave pattern suggests that new issues are reported continuously and inventoried periodically. Therefore, the projects with the patterns indicate they are managing well, and many developers pay attention.