Model-based test case prioritization using selective and even-spread count-based methods with scrutinized ordering criterion

• Main Authors: Mohd. Shafie, M. L., Wan Kadir, W. M. N., Khatibsyarbini, M., Isa, M. A.
• Format: Article
• Language: English
• Published: Public Library of Science 2020
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://eprints.utm.my/id/eprint/88176/1/WanMohdNasir2020_ModelBasedTestCasePrioritization.pdf; http://eprints.utm.my/id/eprint/88176/; http://www.dx.doi.org/10.1371/journal.pone.0229312

Regression testing is crucial in ensuring that modifications made did not introduce any adverse effect on the software being modified. However, regression testing suffers from execution cost and time consumption problems. Test case prioritization (TCP) is one of the techniques used to overcome these issues by re-ordering test cases based on their priorities. Model-based TCP (MB-TCP) is an approach in TCP where the software models are manipulated to perform prioritization. The issue with MB-TCP is that most of the existing approaches do not provide satisfactory faults detection capability. Besides, their granularity of test selection criteria is not very good and this can affect prioritization effectiveness. This study proposes an MB-TCP approach that can improve the faults detection performance of regression testing. It combines the implementation of two existing approaches from the literature while incorporating an additional ordering criterion to boost prioritization efficacy. A detailed empirical study is conducted with the aim to evaluate and compare the performance of the proposed approach with the selected existing approaches from the literature using the average of the percentage of faults detected (APFD) metric. Three web applications were used as the objects of study to obtain the required test suites that contained the tests to be prioritized. From the result obtained, the proposed approach yields the highest APFD values over other existing approaches which are 91%, 86% and 91% respectively for the three web applications. These higher APFD values signify that the proposed approach is very effective in revealing faults early during testing. They also show that the proposed approach can improve the faults detection performance of regression testing.