Discrete cuckoo search algorithms for test case prioritization

Abstract

Regression testing is an essential aspect of the software development lifecycle. As the software evolves, the test suite grows, hence the cost and effort to retest the software. Test case prioritization is one of the mitigation techniques for regression testing. It ranks the test cases to maximize the desired properties, e.g., detecting faults early. The efficiency and effectiveness of test case prioritization techniques can be enhanced using optimization algorithms. Nature-inspired algorithms are gaining more attention due to their easy implementation and quality of the solutions. This paper proposes the discrete cuckoo search algorithm for test case prioritization. The prioritization problem deals with ordering the test cases. Therefore, a new adaptation strategy using asexual genetic reproduction is introduced to convert real numbers into permutation sequences. Furthermore, the cuckoo search algorithm’s effectiveness is extended with the genetic algorithm’s mutation operator to balance the trade-off between exploration and exploitation. An in-depth comparative study on four case studies is conducted between the proposed algorithms, existing state-of-the-art algorithms and baseline approach. Statistical investigation confirms that the proposed hybrid cuckoo search algorithm outperforms the genetic algorithm, particle swarm optimization, ant colony optimization, tree seed algorithm and random search by 4.29%, 5.52%, 8.38%, 2.74% and 10.80%, respectively.