A prioritization method for combinatorial test cases based on small sample tests

Abstract

When the number of parameters or values of systems under test is large, the number of combinatorial test cases increases dramatically. All test cases will take up a lot of time and resources. Prioritization technology can find system faults as early as possible and improve test efficiency. Most of the existing prioritization methods rely on prior knowledge, but in many cases it is difficult to obtain prior knowledge, and random prioritization results are in lower fault detection rates. In this paper, a prioritization method of combinatorial test cases based on small sample tests is proposed. Firstly, weights of value combinations suspected of causing faults are calculated on small sample test cases. Secondly, the number of samples is supplemented by Kmeans++ clustering. Finally, support vector machine model is trained. The predicted results of the model are fused with the weights of value combinations to prioritize combinatorial test cases, and the test cases with the same priority are adjusted according to the distance entropy. Experiments show that the early training of small sample test cases can significantly improve the fault detection speed and the fault detection rate.