DefinePK

Abstract

Software testing is aimed to detect existing faults in a software. The nature of software shows that modification is unavoided. Testing of a modified software is a must to ensure that the software is still free of failures. This process is named as regression testing. Regression testing can be very expensive if all test cases have to be re-tested. To reduce the cost, it is important to prioritize the test case execution to enhance the capability of detecting failures. Test case prioritization is intended to schedule and order the execution of test case based on the certain criteria. In this research, four test case prioritization methods studied emperically are additional branch coverage prioritization, Manhattan distance-based ART (Adaptive Random Testing), additional branch coverage-based with ART, and ART with additional branch coverage-based. Random Testing, as the basic test selection method, is used as a benchmark of the performance of all studied methods. The conducted experiments using two programs as under test program are Replace and Space programs. The experiment results show that all studied methods improve the effectiveness of RT significantly for large program. The used effectiveness measurement is F-measure, the number of test cases executed to detect the first failure. The additional branch coverage-based with ART comes as the best method in terms of F-measure. This method combines the advantage of the additional branch coverage method and the ART. It also reduces the complexity of the additional branch coverage.