Empirical evaluation of test effort efficiency of software GA–based regression test case prioritization strategy

Abstract

GA-based regression test prioritization have ordered test cases by computing fitness value based on the number of affected faults in the coverage information, but most of the researchers use the same severity of faults even if a fault was executed by the previous test case. There have been very little evaluations of the GA-based regression test prioritization, even though there are several studies on GA-based regression test prioritization of object-oriented program (OOP). Most of the evaluations of the previous studies do not consider fault detection efficiency in terms of mutation scores and execution efficiency in terms of execution effort but consider only Average Percentage of the rate of Fault Detection (APFD) metric. The objective of this paper is to integrate the idea of GA with object-oriented programs to aid automated regression test case prioritization of the selected test cases, by proposing a regression test case prioritization strategy for selected test cases of object-oriented programs based on genetic algorithm for efficient OOP regression test case prioritization. This paper proposed an automatic test case prioritization strategy, called HoceDanMafara, and its tool support for Object-Oriented programs. Moreover, a comprehensive empirical study of ten object-oriented programs by the use of mutation analysis was conducted to compare HoceDanMafara and one existing software regression tests prioritization together with non-prioritize and random strategies for regression testing of OOP in term of efficiency of fault detection. The evidence of the efficiency of the proposed strategy are shown in the results of the experiment and statistical tests (p<0.05). The study indicated that the resulting evolutionary tests prioritization produces 27.75% in terms of test effort efficiency compare with randPrior that produces 20.93%, nonPrior produces 14.35% and pSherry produces 20.89%. Therefore, the proposed strategy could be commendable of use as an efficient OOP automatic tests prioritization strategy.