Jean Petrić

Defect prediction has been the subject of a great deal of research over the last two decades. Despite this research it is increasingly clear that defect prediction has not transferred into industrial practice. One of the reasons defect prediction remains a largely academic activity is that there are no defect prediction tools that developers can use during their day-to-day development activities. In this paper we describe the defect prediction tool that we have developed for industrial use. Our ELFF tool seamlessly plugs into the IntelliJ IDE and enables developers to perform regular defect prediction on their Java code. We explain the state-of-art defect prediction that is encapsulated within the ELFF tool and describe our evaluation of ELFF in a large UK telecommunications company.

Background: Test quality is a prerequisite for achieving production system quality. While the concept of quality is multidimensional, most of the effort in testing context has been channelled towards measuring test effectiveness. Objective: While effectiveness of tests is certainly important, we aim to identify a core list of testing principles that also address other quality facets of testing, and to discuss how they can be quantified as indicators of test quality. Method: We have conducted a two-day workshop with our industry partners to come up with a list of relevant principles and best practices expected to result in high quality tests. We then utilised our academic and industrial training materials together with recommendations in practitioner oriented testing books to refine the list. We surveyed existing literature for potential metrics to quantify identified principles. Results: We have identified a list of 15 testing principles to capture the essence of testing goals and best practices from quality perspective. Eight principles do not map to existing test smells and we propose metrics for six of those. Further, we have identified additional potential metrics for the seven principles that partially map to test smells. Conclusion: We provide a core list of testing principles along with a discussion of possible ways to quantify them for assessing goodness of tests. We believe that our work would be useful for practitioners in assessing the quality of their tests from multiple perspectives including but not limited to maintainability, comprehension and simplicity.

Background: Ensemble techniques have gained attention in various scientific fields. Defect prediction researchers have investigated many state-of-the-art ensemble models and concluded that in many cases these outperform standard single classifier techniques. Almost all previous work using ensemble techniques in defect prediction rely on the majority voting scheme for combining prediction outputs, and on the implicit diversity among single classifiers. Aim: Investigate whether defect prediction can be improved using an explicit diversity technique with stacking ensemble, given the fact that different classifiers identify different sets of defects. Method: We used classifiers from four different families and the weighted accuracy diversity (WAD) technique to exploit diversity amongst classifiers. To combine individual predictions, we used the stacking ensemble technique. We used state-of-the-art knowledge in software defect prediction to build our ensemble models, and tested their prediction abilities against 8 publicly available data sets. Conclusion: The results show performance improvement using stacking ensembles compared to other defect prediction models. Diversity amongst classifiers used for building ensembles is essential to achieving these performance improvements.

Background: The NASA datasets have previously been used extensively in studies of software defects. In 2013 Shepperd et al. presented an essential set of rules for removing erroneous data from the NASA datasets making this data more reliable to use. Objective: We have now found additional rules necessary for removing problematic data which were not identified by Shepperd et al. Results: In this paper, we demonstrate the level of erroneous data still present even after cleaning using Shepperd et al.'s rules and apply our new rules to remove this erroneous data. Conclusion: Even after systematic data cleaning of the NASA MDP datasets, we found new erroneous data. Data quality should always be explicitly considered by researchers before use.

BACKGROUND -- During the last 10 years hundreds of different defect prediction models have been published. The performance of the classifiers used in these models is reported to be similar with models rarely performing above the predictive performance ceiling of about 80% recall. OBJECTIVE -- We investigate the individual defects that four classifiers predict and analyse the level of prediction uncertainty produced by these classifiers. METHOD -- We perform a sensitivity analysis to compare the performance of Random Forest, Naïve Bayes, RPart and SVM classifiers when predicting defects in 12 NASA data sets. The defect predictions that each classifier makes is captured in a confusion matrix and the prediction uncertainty is compared against different classifiers. RESULTS -- Despite similar predictive performance values for these four classifiers, each detects different sets of defects. Some classifiers are more consistent in predicting defects than others. CONCLUSIONS -- Our results confirm that a unique sub-set of defects can be detected by specific classifiers. However, while some classifiers are consistent in the predictions they make, other classifiers vary in their predictions. Classifier ensembles with decision making strategies not based on majority voting are likely to perform best.