Evolved regression test suite selection using BCO and GA and empirical comparison with ACO

Abstract

Regression testing is a maintenance activity that is performed to ensure the validity of modified software. The activity takes a lot of time to run the entire test suite and is very expensive. Thus it becomes a necessity to choose the minimum set of test cases with the ability to cover all the faults in minimum time. A lot of techniques have already been developed and proved to be very effective in reduction of test suite. The paper presents a new improved modified technique based on Bee Colony Optimization and Genetic Algorithm that makes use of permutations/combinations to generate a new set of test cases. The proposed modified technique has been empirically validated on 17 sample programs for its near 90 % correctness and an average 80 % execution time reduction capability. Also, the developed technique has been compared with the existing technique for test case selection using ACO. The comparison proves the superiority of the developed technique against the existing one in majority of the programs with some exceptions. In addition to this, the results have been analyzed based on the language of the programs under test and the type of desired result. The comparison between the tool MHBG_TCS (developed for the technique proposed in this paper) and the tool ACO_TCSP (existing tool) yielded superiority of the new technique in general. All the results prove the validity of our technique and inspire us to work further on this technique.