Influence of Structural Complexity on Fault Persistence

M. Pighin and A. Marzona (Italy)


Testing, Maintenance, Metrics, Reliability


In our previous works we empirically verified the relationship between a high number of faults at a source file's first release and its fault proneness through the following releases. We called "faulty files" files collecting, at their first release, a number of faults higher than a given value. In this paper we investigate the causes of these fault proneness, observing the structure of faulty files, measuring them using both cyclomatic complexity index and a risk predictive structural metric. Following these measures, faulty files are structurally more complex than other files, and their fault proneness is substantially due to structural complexity. Faulty files can become more reliable lowering their structural complexity.

Important Links:

Go Back