Abstract:An effective system regression testing for consecutive releases of very large software systems, such as modern telecommunications systems, depends considerably on the selection of test cases for execution. Classification models can classify, early in the test planning phase, those test cases that are likely to detect faults in the upcoming regression test. Due to the high uncertainties in regression test, classification models based on fuzzy logic are very useful. Recently, methods have been proposed for automatically generating fuzzy if-then rules by applying complicated rule generation procedures to numerical data. In this research, we introduce and demonstrate a new rule-based fuzzy classification (RBFC) modeling approach as a method for identifying high effective test cases. The modeling approach, based on test case metrics and the proposed rule generation technique, is applied to extracting fuzzy rules from numerical data. In addition, it also provides a convenient way to modify rules according to the costs of different misclassification errors. We illustrate our modeling technique with a case study of large-scale industrial software systems and the results showed that test effectiveness and efficiency was significantly improved.

A Novel Fuzzy Classification to Enhance Software Regression Testing