Abstract: To reduce the loss of fault detection capability during the test suite reduction process and to minimize the negative impact of the coincidental correctness on test adequacy. A test suite reduction method is proposed based on hierarchical clustering, which uses Euclidean distance to calculate the similarity of test cases and derives the dissimilarity matrix as the input for clustering. In addition, based on the property that mutation testing can improve the fault detection ability of test suite, a formula for calculating the probability of coincidental correctness is given. The reduced test suite are obtained in the resultant clusters using the mutation testing-based test case selection algorithm. The experimental results show that the proposed method reduces the loss of fault detection capability by 9.8% and 12% compared to adequate and other inadequate test suite reduction methods, respectively.

Key words: hierarchical clustering, test suite reduction, mutation testing, coincidental correctness, fault detection capability