Do different cross‐project defect prediction methods identify the same defective modules?

Chen, Xiang; Mu, Yanzhou; Qu, Yubin; Ni, Chuanfa; Liu, Meng; He, Tong; Liu, Shangqing

doi:10.1002/smr.2234

Cited by 13 publications

(9 citation statements)

References 95 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We further use Scott-Knott test [23] to rank our proposed method ALTRA and all the seven baselines. Scott-Knott test was recommended by Ghotra et al [16] and then widely used in previous empirical studies for software defect prediction [8]- [11], [50]. This method performs the grouping process in a recursive way.…”

Section: Results Analysis a Results Analysis For Rq1 1) Rq1: Can Oumentioning

confidence: 99%

See 1 more Smart Citation

ALTRA: Cross-Project Software Defect Prediction via Active Learning and Tradaboost

et al. 2020

Self Cite

View full text Add to dashboard Cite

Cross-project defect prediction (CPDP) methods can be used when the target project is a new project or lacks enough labeled program modules. In these new target projects, we can easily extract and then measure these modules with software measurement tools. However, labeling these program modules is time-consuming, error-prone and requires professional domain knowledge. Moreover, directly using labeled modules in the other projects (i.e., the source projects) can not achieve satisfactory performance due to the large data distribution difference in most cases. In this article, to our best knowledge, we are the first to propose a novel method ALTRA, which can utilize both active learning and TrAdaBoost to alleviate this issue. In particular, we firstly use Burak filter to select similar labeled modules from the source project after analyzing the unlabeled modules in the target project. Then we use active learning to choose representative unlabeled modules from the target project and ask experts to label the type (i.e., defective or non-defective) of these modules. Later, we use TrAdaBoost to determine the weights of labeled modules in the source project and the target project, and then construct the model via weighted support vector machine. After selecting a small number of modules (i.e., only 5% modules) in the target project, we terminate the method ALTRA and return the final constructed model. To show the effectiveness of our proposed method ALTRA, we choose 10 large-scale open-source projects from different application domains. In terms of both F1 and AUC performance indicators, we find ALTRA can perform significantly better than seven state-of-the-art CPDP baselines. Moreover, we also show that the usage of Burak filter, the uncertainty active learning strategy, the class imbalanced learning method and TrAdaBoost are competitive in our proposed method ALTRA.

show abstract

Section: Results Analysis a Results Analysis For Rq1 1) Rq1: Can Oumentioning

confidence: 99%

“…To better rank our proposed method ALTRA and all the CPDP baselines in terms of a specific performance indicator, we use Scott-Knott test [23]. Since recent studies [8]- [11], [50] have suggested that using the Scott-Knott test is superior to some post hoc tests (e.g., Friedman-Nemenyi test).…”

Section: Threats To Conclusion Validitymentioning

confidence: 99%

ALTRA: Cross-Project Software Defect Prediction via Active Learning and Tradaboost

et al. 2020

Self Cite

View full text Add to dashboard Cite

show abstract

“…Is it useful for just starting phase of the new project or it can also work when the project is old? Chen et al [13] carried out a comparative study between eight supervised and four unsupervised defect prediction models. The main idea was to check if different models predict the same defective modules.…”

Section: Previous Work 21 Just-in-time Software Defect Predictionmentioning

confidence: 99%

Defect Prediction Using Akaike and Bayesian Information Criterion

Albahli¹,

Yar²

2022

Computer Systems Science and Engineering

View full text Add to dashboard Cite

“…To show whether there exists a statistical difference between our proposed IETCR strategy with baselines in terms of the EXAM score metric, we use the Wilcoxon signed-rank test. Since this kind of statistical test method has been widely used in previous studies [21], [49]- [52].…”

Section: Threats To Validitymentioning

confidence: 99%

IETCR: An Information Entropy Based Test Case Reduction Strategy for Mutation-Based Fault Localization

Wang

et al. 2020

IEEE Access

Self Cite

View full text Add to dashboard Cite

Mutation-based fault localization (MBFL) is a recently proposed technique with the advantage of high fault localization accuracy. However, such a mutation analysis based technique is difficult to be accepted by industry due to its huge computational cost on mutation analysis. There are three ways to improve MBFL's efficiency, which are reducing the number of mutants, optimizing the mutants' execution process, and reducing the number of test cases. The former two ways have been mainly studied and shown promising results, but for the latter way, the related studies are limited since this kind of method will reduce the precision of MBFL. In this paper, we mainly focus on the latter way and propose an information entropy based test case reduction (IETCR) strategy for MBFL. In particular, we first calculate the entropy change of test cases and select a proportion of them according to their value. Then we use a reduced test suite to execute mutants. To show the effectiveness of the IETCR strategy, we choose six real-world programs with faulty versions. In terms of mutation reduction rate, we find MBFL with the IETCR strategy can reduce 56.3%∼88.3% cost while keeping almost the same fault localization accuracy when compared to the original MBFL without test case reduction. Moreover, we use Wilcoxon signed-rank test for statistical analysis, which shows that there is no statistically significant difference between MBFL with IETCR strategy and the original MBFL. INDEX TERMS Software fault localization, Mutation based fault localization, Information entropy, Test case reduction Recent studies [11], [12] have shown that mutation-based fault localization (MBFL) techniques outperform state-ofthe-art SBFL techniques in terms of the fault localization

show abstract

Do different cross‐project defect prediction methods identify the same defective modules?

Cited by 13 publications

References 95 publications

ALTRA: Cross-Project Software Defect Prediction via Active Learning and Tradaboost

ALTRA: Cross-Project Software Defect Prediction via Active Learning and Tradaboost

Defect Prediction Using Akaike and Bayesian Information Criterion

IETCR: An Information Entropy Based Test Case Reduction Strategy for Mutation-Based Fault Localization

Contact Info

Product

Resources

About