Relationship Between Design and Defects for Software in Evolution

In software development, one of the main problems is predicting defects during the software product life cycle. In our analysis, we focused on that problem and investigated whether it can be tackled as early as in the design stage of evolving software. In the design stage of software life cycle, most of the effort is invested in emergence of a module. Then, with each upgrade of this module, it is possible to break concepts of good design. Numerous changes are introduced, and these changes can also cause defects. From this standpoint, we would like to find the design metrics that are critical to the appearance of defects. Product metrics have been successfully used to build prediction models [Basili et al. 1996], so a great number of such metrics is investigated in this study. When developing a software product, it is essential to plan the necessary actions needed to upgrade a certain functionality. Often, through this process, developers need to modify, delete or add new lines of program code. The amount of program code that goes through these actions is called code churn. Developers should strive to minimize the quantity of program lines that will be modified, thus achieving a smaller code churn ratio [Munson and Elbaum 1998]. After the development phase, the software