Complexity-based Prediction of Faults Number for Software Modules Ranking Before Testing: Technique and Case Study
Svitlana Yaremchuk, Vyacheslav Serhiiovych Kharchenko · 2016
The proposed method is based on estimating software modules’ complexity by means of metrics. Indices of source code complexity are the in- put data for the method. Ranged selection software modules with faults form the output data for the method. The verification of the method has been per- formed on the basis of representative selection of experimental data for five systems. An appropriate coefficient is proposed and calculated for assessment of the method’s efficiency. It shows the correlation between a part of revealed faults to a part of verified modules. The efficiency coefficient value increases on condition of increase in quantity of applied metrics. The proposed method enables to reach the required reliability by means of a choice of certain part of modules for identification as it is possible for the most part of faults. The method provides a software reliability increase in limited resources conditions.