On the value of learning from defect dense components for software defect prediction

Abstract

BACKGROUND: Defect predictors learned from static code measures can isolate code modules with a higher than usual probability of defects. AIMS: To improve those learners by focusing on the defect-rich portions of the training sets. METHOD: Defect data CM1, KC1, MC1, PC1, PC3 was separated into components. A subset of the projects (selected at random) were set aside for testing. Training sets were generated for a NaiveBayes classifier in two ways. In sample the dense treatment, the components with higher than the median number of defective modules were used for training. In the standard treatment, modules from any component were used for training. Both samples were run against the test set and evaluated using recall, probability of false alarm, and precision. In addition, under sampling and over sampling was performed on the defect data. Each method was repeated in a 10-by-10 cross-validation experiment. RESULTS: Prediction models learned from defect dense components out-performed standard method, under sampling, as well as over sampling. In statistical rankings based on recall, probability of false alarm, and precision, models learned from dense components won 4-5 times more often than any other method, and also lost the least amount of times. CONCLUSIONS: Given training data where most of the defects exist in small numbers of components, better defect predictors can be trained from the defect dense components.

Can Reused Components Provide Lead to Future Defective Components in Smart Grid Applications?

“…In practical use, defect predictors are useful for adjusting quality assurance (QA) budgets to focus on blind spots especially for critical infrastructure [25]. In addition, this approach can be used to indicate early warning signs of any possible defect for dependent components.…”

“…It is common to focus QA efforts on the most important parts of the system, however, as noted by Zhang et al [25], a perceived less important system in a SoS can be ignored for quality assurance, thus leaving blind spots in the system that can compromise such other high assurance systems. A case cited involved a surprise anomaly that resulted into modification of ground software at NASA which is a perceived less important part when compared to the on-board guidance system.…”

“…To further understand the distribution of the defect data, we determined the average of defects produced by the defective components. As pointed out in Zhang et al [25], learning from defect dense components is useful knowledge in defect prediction modeling. Given that, a defective component can have a certain number of defects in a particular release that is greater or equal to one.…”

Software Defect Prediction Model Research for Network and Cloud Software Development

“…In order to make the software company to produce high quality products, to avoid or prevent software defects caused by adverse effects on the application system, the researchers in the field of software engineering has done a lot of research, reason and process perspectives arising from the software defects, to explore the solution of software defects [1][2][3][4] , including the occurrence and trend through the development and application of various software defect management tools to effectively manage software defects, these research results help software enterprises and developers, testers should try to avoid the defects in the software engineering development process, to a great extent. Foreign researchers in basic research work, proposed a variety of software defect prediction model, including: the scale of software defect prediction technology, complexity of software defect prediction technology, forecast technology, based on machine learning metrics based on multidimensional software defect prediction technology based on technology.…”

A Comparison of Different Defect Measures to Identify Defect-Prone Components

“…However, existing defect models are largely constructed to predict based on number of defects (e.g. [2][3][4][5][6][7][8][9]). The study by Adams [10] showed that removing large number of defects may have a trivial effect on reliability.…”