Hybrid search-based and string similarity-based prioritization technique for regression testing

• Main Author: Kamil Riadz, Muhammad Irsyad
• Format: Thesis
• Language: English
• Published: 2022
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://eprints.utm.my/102994/1/MuhammadIrsyadMSC2022.pdf.pdf; http://eprints.utm.my/102994/; http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:150718

Testers have popularly used regression testing in detecting errors encountered after changes were made. Numerous techniques were introduced in maximizing average percentage fault detection (APFD). Based on recent studies, test case prioritization (TCP) technique can give the highest APFD score. However, each approach used in TCP has limitations such as high execution cost and lack of information. Approaches that can cover more than one variable of test suite remained unseen. Thus, there is a need for a hybrid TCP technique to be developed to search for the best test plan that gives a good APFD score while having a good coverage of test cases relevant to the cost execution. Ordering the test cases based on the string similarity is one of the conventional approaches used by researchers. With the usage of string similarity, the study can gain more information regarding the test suite. This study aims to maximize the high rate of fault detection while reducing cost by decreasing the number of test cases. In this research, two TCP techniques which are string similarity-based and search-based were hybridized to form a new hybrid TCP technique and applied with Test Case Selection using weight-based to consider more variables during regression. The whole process begins by calculating string similarity for TCP with an enhanced Jaro-Winkler, then prioritizing test cases using a search-based approach with a genetic algorithm based on fault revealing. Each process generates a test plan, and those test plans will be merged and selected to form a new test plan. The selection process is structured using a weight-based approach. The experimental result showed that the final test plan produced second highest APFD with 89.60%, covers 74.10% of test case coverage and 82% of APFD, covering 77.55% of test case coverage in Siemens dataset and Smart Wheelchair System (SWS) case study. In conclusion, the proposed technique has benefited all approaches applied by getting a good APFD and coverage score. Thus, the proposed technique has proven to be cost-effective, as the APFD and coverage score are significant as the size of test suite decreases.