A novel approach to multiple criteria based test case prioritization
Robeala Abid, Aamer Nadeem · 2017
When software is modified, it is retested to ensure that no new faults have been introduced in the previously tested code and it still works correctly. Such testing is known as regression testing. The cost of regression testing is high because the original program has large number of test cases. It is not feasible to execute all test cases for regression testing. Test suite minimization, test case selection and test case prioritization are cost commonly used techniques in regression testing to reduce the cost of regression testing. While test suite minimization and test case selection techniques select a subset of test cases, test case prioritization does not eliminate any test case, it only orders the test cases with the objective of increasing the fault detection rate. Prioritization is usually preferred over other two approaches because it does not involve the risk of losing useful test cases. Prioritization techniques assign priority to each test case on the basis of some coverage criteria. A number of different single criterion and multiple criteria based prioritization techniques have been proposed in the literature. Multiple criteria based prioritization techniques perform better than single criterion based prioritization techniques. The existing multiple criteria based prioritization techniques combine the criteria in such a way that “Additional” strategy cannot be applied on them. In this paper, we propose a new multiple criteria based test case prioritization algorithm that considers two criteria to prioritize test cases using “Additional” strategy. One criterion is considered as primary and other is considered as secondary. Primary criterion is used to prioritize the test cases whereas secondary criterion is used to break the tie among test cases when two or more test cases provide equal coverage of entities of first criterion. Our proposed multiple criteria based prioritization algorithm performs better than the existing prioritization techniques.