An Application of Association Rule Mining to Unit Test Selection
View/ Open
Abstract
Appropriate selection of unit tests during the software development process is vital when many unit tests exist. The developer may be unfamiliar with some tests and non-obvious relationships between application code and test code may exist. Poor test selection may lead to defects. This is especially true when the application is large and many developers are involved. By the application of association rule mining to the unit test selection process and by comparison with extant selection techniques, we will provide a quantitative analysis of the benefits of heuristic and its limit to development where process patterns are stable.