A Genetic Algorithm-Based Approach for Test Case Prioritization

Abstract

Software maintenance is the most important and expensive activity in the process of Software Development Life Cycle (SDLC). In the maintenance stage, after modification, the software goes through verification and validation. Regression testing is performed at the maintenance stage of SDLC to ensure that the old functionalities are working perfectly. Test case prioritization, which is about making a sequence of the test case, is one of the important parts of regression testing. Test case prioritization(TCP), which is a class of NP-hard problem, can have a better solution using soft computing approach as per no-free-lunch theorem [1]. The same no-free-lunch theorem states that a soft computing approach yields a case-specific result. In this article, we propose a novel genetic algorithm approach to solve the TCP problem. The proposed algorithm is experimentally compared with random technique. For the experiment three-benchmark program from the software artifact infrastructure repository is selected. From the experiment, it was found that the Average Percentage of Fault Detection (APFD) of the GA technique gives a better result than random technique. Also, an experiment was conducted to record the execution time with different numbers of generations. The results confirm that the result is not directly dependent on the number of generation.