CMHFL: A new Fault Localization technique based on Cochran–Mantel–Haenszel method

As modern software systems are large and complex, manually debugging software has become excessively expensive, time-consuming, boring, and error-prone. We present a modified Cochran–Mantel–Haenszel statistical method that utilizes test results along with statement coverage information in order to compute the suspiciousness score ( likelihood of containing a fault) of each statement and generate a ranked list of statements. Results show that on an average, our proposed approach outperforms in contrast to other Fault Localisation (FL) techniques, such as Zoltar, Dstar, Ochiai, Barinel, and Ample by 38.994%.