A New Approach Based on Software Metrics to Improve the Effectiveness of Regression Testing

Test case prioritization has been often used to alleviate the costs associated with software regression testing. Current techniques have attempted to estimate the fault exposing potential of test cases using code coverage information and rank them using a heuristic. However, studies show that coverage does not strongly correlate with the effectiveness and fault exposing potential of test cases. Relying on the results of studies that demonstrated the effectiveness of code metrics in fault prediction, we speculate that code metric information can be leveraged to design a new effective technique for test case prioritization. Based on our hypothesis, in this paper, a new prioritization technique is proposed that works based on data fusion on code complexity metrics. The novelty of our technique lies in its original viewpoint to estimate fault exposing potential of test cases in prioritization. To evaluate the proposed technique, we have conducted experiments on faulty versions of seven Java benchmarks. In the experiments, we often observed at least 70% performance in prioritization measured in terms of average percentage of fault detection, which validates our hypothesis.