Towards building a universal defect prediction model

Abstract

To predict files with defects, a suitable prediction model must be built for a software project from either itself (withinproject) or other projects (cross-project). A universal defect prediction model that is built from the entire set of diverse projects would relieve the need for building models for an individual project. A universal model could also be interpreted as a basic relationship between software metrics and defects. However, the variations in the distribution of predictors pose a formidable obstacle to build a universal model. Such variations exist among projects with different context factors (e.g., size and programming language). To overcome this challenge, we propose context-aware rank transformations for predictors. We cluster projects based on the similarity of the distribution of 26 predictors, and derive the rank transformations using quantiles of predictors for a cluster. We then fit the universal model on the transformed data of 1,398 open source projects hosted on SourceForge and GoogleCode. Adding context factors to the universal model improves the predictive power. The universal model obtains prediction performance comparable to the within-project models and yields similar results when applied on five external projects (one Apache and four Eclipse projects). These results suggest that a universal defect prediction model may be an achievable goal.

An empirical study on predicting defect numbers

“…For example, Turhan et al [26] applied a nearest neighbor filtering technique to filter out those irrelevant project data in the setting of CPDP, leading to a better prediction performance. More discusses on the comparison between WPDP and CPDP please refer to [4,10,27,28]. Unfortunately, very few prior studies paid attention to the issue in question in CPDP settings.…”

“…So, we should take into consideration various factors (rather than just accuracy) when applying them to different types of actual projects with limited resources, which is required to make an optimal (or near-optimal) tradeoff among generality, performance and construction cost. That is, we want to find one or more appropriate regression models that can be used in different scenarios, because the previous studies about defect-proneness prediction have showed that the classifiers which are simple and easy to use tend to perform well in both within-and cross-project scenarios [10,27]. In particular, is this still practicable for defect numbers prediction?…”

CloCom: Mining existing source code for automatic comment generation

“…• Remove code comments that contain any of the commonly used terms in defect prediction [20]. bug, fix, error, issue, crash, problem, fail, defect, patch …”

The RISCOSS Platform for Risk Management in Open Source Software Adoption

“…In the area of defect prediction for quality improvement, Peters et al [5] introduce guidelines to be used in the building of software quality predictors in case of scarcity of data while D'Ambros et al present a comparison between the different prediction approaches [6]. Zhang et al Platform to obtain an objective value of the software development process quality present in [7] a study for the specification of a universal defect predictor. Gamalielsson et al [8] define the health of an open source ecosystem as an important decision factor when considering the adoption of an OSS component.…”