Author(s)

Wenke Yang, Yuan Wei, Fu Yin, Zhengyi Yang

Corresponding Author

Zhengyi Yang

ABSTRACT

Open source software (OSS) has played an essential role in this era. Many companies and researches rely on open source software. However, not all open source projects attract enough developers and get improved regularly. In this paper, to find out factors on contributing to OSS's; we analysed projects with large team size from the TravisTorrent dataset. Machine learning based feature selection and Pearson's correlation analysis are applied to the data and the results show that test density and assert density have strong negative impacts on attracting new contributors. To attract more members in OSS, we suggest that one may pay more attention on production files other than focusing too much on tests. Future researchers may also apply the methodologies used in this paper to explore other factors and/or apply it on different datasets.

KEYWORDS

Open source, Travistorrent, Pearson correlation, feature selection, data analysis, machine learning