Distribution of Test Data for Regression Test

Description

Regression testing is time-consuming, and discarding some test cases may compromise fault detection capability. Test case prioritization is an alternative. Many existing test case prioritization techniques use the white-box code coverage as a surrogate measure to guide their permutation process. Nonetheless, achieving high rate of code coverage does not necessarily entail high rate of fault detection for every program. In this project, we propose to investigate test input data prioritization strategies. Our study aims to answer the following research questions: (1) What are the desirable characteristics of distributions of input characteristics of effectively permutated test suite? (3) Is it feasible to develop an effective and efficient test data oriented strategy that exhibits these desirable characteristics? The result of this project both deepens our understanding on regression testing and improve the state-of- the-art.