Design and code coupling assessment based on defects prediction. Part 1

The article discusses an application of code metrics at objectoriented software design. Code metrics give an additional method to avoid errors except the obvious ones like thorough requirements, design, programming, testing, and consumer’s feedback. Software metrics try to collect values and measurements from the software and predict possible current or future problems. This paper includes the development, analysis and evaluation of several software code metrics. The paper also investigates how could coupling metrics be utilized as early indicators of fault proneness. A tool is developed to parse through code projects and automatically collect those metrics. A case study of Scarab project is selected to evaluate coupling metrics ability to predict fault proneness. Results showed that the value of the evaluated metrics can vary in terms of their ability to judge the software design and fault proneness. Results showed also that CBO, RFC, MPC and ICP have more correlation with reported bugs in comparison with other collected and evaluated coupling metrics.