KS-TCP: An Efficient Test Case Prioritization Approach based on K-medoids and Similarity

Abstract

Test case prioritization (TCP) tries to find an optimal execution sequence by adjusting test cases that need to be executed. Traditional techniques rely on code coverage information to achieve effective results, but they need access to historical execution information. The string distance-based test case prioritization (SD-TCP) can avoid these limitations through only using the test cases themselves for sorting, but it is sensitive to extreme test cases and inefficient. To overcome these problems, we propose a test case prioritization method based on K-medoids and Similarity (KS-TCP). The proposed KS-TCP approach considers sorting a set of test cases rather than individual test case to effectively avoid the effect of extreme test cases, it uses cluster analysis and greedy strategy to divide the subsets and compose the final execution sequence by polling. Extensive experimental results show that the proposed KS-TCP approach has a higher APFD value compared to Random Prioritization (RP) and SD-TCP, and it also outperforms SD-TCP in terms of better time efficiency on test case prioritization.