Improving open source software process quality based on defect data mining

Abstract

Open Source Software (OSS) project managers often need to observe project key indicators, e.g., how much efforts are needed to finish certain tasks, to assess and improve project and product quality, e.g., by analyzing defect data from OSS project developer activities. Previous work was based on analyzing defect data of OSS projects by using correlation analysis approach for defect prediction on a combination of product and process metrics. However, this correlation analysis is focusing on the relationship between two variables without exploring the characterization of that relationship. We propose an observation framework that explores the relationship of OSS defect metrics by using data mining approach (heuristics mining algorithm). Major results show that our framework can support OSS project managers in observing project key indicators, e.g., by checking conformance between the designed and actual process models. © 2012 Springer-Verlag Berlin Heidelberg.