Test case prioritization using multi objective particle swarm optimizer

Abstract

The goal of regression testing is to validate the modified software. Due to the resource and time constraints, it becomes necessary to develop techniques to minimize existing test suites by eliminating redundant test cases and prioritizing them. This paper proposes a 3-phase approach to solve test case prioritization. In the first phase, we are removing redundant test cases by simple matrix operation. In the second phase, test cases are selected from the test suite such that selected test cases represent the minimal set which covers all faults and also at the minimum execution time. For this phase, we are using multi objective particle swarm optimization (MOPSO) which optimizes fault coverage and execution time. In the third phase, we allocate priority to test cases obtained from the second phase. Priority is obtained by calculating the ratio of fault coverage to the execution time of test cases, higher the value of the ratio higher will be the priority and the test cases which are not selected in phase 2 are added to the test suite in sequential order. We have also performed experimental analysis based on maximum fault coverage and minimum execution time. The proposed MOPSO approach is compared with other prioritization techniques such as No Ordering, Reverse Ordering and Random Ordering by calculating Average Percentage of fault detected (APFD) for each technique and it can be concluded that the proposed approach outperformed all techniques mentioned above.