Adaptive Weighted Sum Bi-objective Bat for Regression Testing Optimization
Abstract: Regression testing is a type of testing carried out during the software maintenance phase, to confirm the validity of a software system after any modifications. However, regression testing is expensive, and sometimes it cannot be carried out within the testing budget, due to the large size of a test suite. In order to reduce regression testing cost, the test suite should be reduced without losing its efficiency in terms of pre-defined criteria such as its capability of fault detection; this problem is known as test suite reduction problem (TSR). In this paper, the TSR problem was formulated as a bi-objective optimization problem using an adaptive-weighted (AW) sum method. Then an adapted binary Bat algorithm (AW-ABBA) was utilized to search for a Pareto-optimal set of solutions; allowing the decision-maker under various circumstances to choose the best solution from the proposed set. The efficacy of the AW-ABBA was assessed using three metrics, Cardinality ratio, \({IGD}^{+}\) and Diversity, over five test suites of different sizes. Experimental results showed that the AW-ABBA was able to efficiently approximate a reference Pareto-front.
Loading