A comparative study of Bat and Cuckoo search algorithm for regression test case selection

Abstract

Enhancing the software by either adding new functionality or deleting some obsolete capability or fixing the errors is called software maintenance. As a result, the software may function improperly or unchanged parts of the software may be adversely affected. Testing carried out to validate that no new errors have been introduced during maintenance activity is called Regression Testing. It is acknowledged to be an expensive activity and may account for around 60–70% of the total software life cycle cost. Reducing the cost of regression testing is therefore of vital importance and has the caliber to reduce the cost of maintenance also. This paper evaluates the performance of two metaheuristic algorithms-Bat Algorithm and Cuckoo Search Algorithm for selecting test cases. Factors that we have considered for performance evaluation are the number of faults detected and the execution time. The domain of study is the flex object from the Benchmark repository — Software Artifact and Infrastructure Repository. Extensive experiments have been conducted to collect and analyze the results. A Statistical test, F-test has also been conducted to validate the research hypothesis. Results indicate that the Cuckoo Search Algorithms perform a little better than Bat Algorithm.