Related and Future Work

The work of Mattsson and Bosch is the basis for our approach to identifying the software evolution through examining historical information. We extended their approach in some ways to fit the specific needs of our research question. Mattsson and Bosch extended the original approach of Gall et al. in the sense of smaller granularity of the entities examined [Gall+1997]. Our method extends the approach of Mattsson and Bosch in the sense that we examine the software evolution over smaller periods of time to better fit the incremental and iterative development in small steps.

An empirical study by Lindvall and Sandahl shows that software developers are not so good at predicting, based on the requirements specification, how many and which classes will be changed [Lindvall+1998]. In the context of XP, the idea of a requirements document is omitted completely in favor of user stories, which contain only the next most important requirement for the evolution of the system [Beck2000].

Simon and Steinbrückner have analyzed JWAM 1.5 with their high-quality metrics tool. They are working on an analysis of a more recent version of JWAM to study the effects of their first recommendations on the quality of the framework in that version [Simon+2001].

Our analysis concentrated on public methods given by the method we base our work on. Experience shows, however, that refactoring is applied to private methods in many cases. On the other hand, our analysis didn't take into account the correlation between requirements changes and defect rates, which could reveal other insights into the development process.