Enhanced non-dominated sorting genetic algorithm for test case optimization

Abstract

Due to inevitable software changes, regression testing has become a crucial phase in software development process. Many software testers and researchers agreed that regression testing process consumes more time and cost during software development. Test case optimization has become one of the best solutions to overcome problems in regression testing. Test case optimization is focusing on reducing number of test cases in the test suite that may reduce the overall testing time, cost and effort of software testers. It considers multiple objectives and provides several numbers of optimal solution based on objectives of the testing. Therefore, this research aims at developing an alternative solution of test case optimization technique using NSGA II with fitness scaling as an additional function. Fitness scaling function is applied in NSGA II to eliminate pre-mature convergence among set of solution in the evolution of offspring in NSGA II which may produce more efficient fitness value. This research focuses on regression testing optimization by implementing weight of test cases and fault detection rate per test case as its objective function for optimization purposes. The proposed technique is applied to the GUI-based testing case study. The result shows that Pareto front produced by enhanced NSGA II give more wider set of solution that contains more alternatives and provide better trade-off among solutions. The evaluation shows that enhanced NSGA II perform better compared to conventional NSGA II by increasing the percentage of the reduced test cases with 25% and yield lower fault detection loss with 1.64% which indicating that set of reduced test cases using enhanced NSGA II is able to maintain the fault detection capability in the system under test.