Abstract:

Testing is one of the most expensive but essential software development activity that helps the software professionals to deliver quality software. The quality of software is judged on the basis of number of faults lying dormant inside the software. Software systems are developed by integrating various independent modules. These modules are neither equally important nor do they contain an equal amount of faults and may be categorized as fault-prone (FP) or not fault-prone (NFP) depending on the number of fault present in the module. FP modules may require more testing than NFP modules because of its likelihood of containing more faults. Also, modules either NFP or FP may not have the equal fault-prone degree and therefore testing resources should be allocated on the basis of its fault-prone degree. Therefore, it is desirable to rank these FP modules on the basis of its fault prone degree. Ranking helps software professionals to prioritize their testing action.
This paper presents a new approach of early software quality prediction and ranking. Quality prediction is done by classifying software modules as FP or NFP. Furthermore, modules are ranked using software metrics and fuzzy ordering algorithm on the basis of their degree of fault proneness. Ranking of fault-prone module along with classification found to be a new approach to help in prioritizing and allocating test resources to the respective software modules. The model accuracy is validated through KC2 dataset. The results observed are found promising, when compared with some of the earlier models.