Software development methodologies, or process models, attempt to describe the steps that should be followed along the way from conception to deployment of software. There are traditional approaches that focus on the sequence of discrete and well-defined steps, like the Waterfall model, where communication channels are realized by passing documents, and others, like the Agile model, which emphasizes on the need for flexibility and constant, direct communication between team members. These newer models are very popular with software teams of varying sizes. Due to the importance and the means of communications described by these models, it is desired to recruit people that possess both technical and communicational skills. The problem, though, that arises when looking for people like these, lies in the difficulty of assessing these skills. Within the context of this project we focus on this issue. To do so we employ data mining techniques for identifying different team roles and also assess the activities of team members within the software development and operations process. The implemented system draws user activity data from the GitHub web platform and uses them as input to cluster team members. This way we attempt to provide insight into the different team member roles that appear in open source projects, like the ones at GitHub, and the performance of the users that act under these roles. After extensive experimentation with different combinations of datasets and evaluation features, the results that are presented as final are considered to offer critical insight into those matters.
Full report @ https://github.com/Zaf97/Software_Engineering_Role_Recognition_using_ML/blob/master/Report.pdf