Discrete and combinatorial gravitational search algorithms for test case prioritization and minimization

Abstract

Regression testing is an essential but expensive activity to re-execute all the test cases every time the software updates. Test case prioritization and minimization reduces the cost and efforts required for retesting by prioritizing the test cases based on their importance and minimizing the redundancy. Optimization approaches further enhance the effectiveness of these techniques. In this paper, a discrete and combinatorial gravitational search algorithm is proposed to solve the test case prioritization and minimization problems. Furthermore, an improved version is developed using the chaotic map to update the gravitational constant. The proposed algorithms are compared with the most commonly used algorithm, i.e., genetic algorithm. Three subject programs of varying sizes are used for evaluation. Simulation results prove that the proposed algorithms are more efficient and effective than the genetic algorithm for test case prioritization and minimization. Statistical representation via boxplots of APFD and interval plots of minimized suite size performance metrics, confirms that the improved gravitational search algorithm with chaotic gravitational constant has a more squeezed distribution than the standard gravitational search algorithm.