A regression test case prioritization technique targeting ‘hard to detect’ faults

We propose a novel regression test case prioritization technique targeting to detect ‘hard to detect’ faults. A ‘hard to detect fault’ is a fault that is detected by only one test case. In our technique, we first seed a large number of bugs into a program to create mutants. Each mutant is executed with the test suite and their execution results are recorded in a fault matrix. Using the fault matrix, we first prioritize the test cases based on their ‘hard to detect’ values. Our technique assigns higher priority to those test cases which reveal bugs in a program-component that are hard to expose. Subsequently, the remaining test cases are prioritized based on their fault detection capability. Our experimental results show that on an average our proposed TCP technique performs 43.82% better than existing TCP techniques.