Weighting for Combinatorial Testing by Bayesian Inference

Authors:
  • ECEun-Hye Choi
    Eun-Hye Choi
  • TFTsuyoshi Fujiwara
    Tsuyoshi Fujiwara
  • OMOsamu Mizuno
    Osamu Mizuno

Abstract
Combinatorial testing (CT) is a widely-used technique to detect system interaction failures. To improve the test effectiveness of CT, prioritized combinatorial testing inputs priority weights of parameter-values, and generates combinatorial test suites based on the weights. This paper proposes a method to automatically determine the weights of parameter-values by Bayesian inference using previous testing results. Using two open source projects, we evaluate the fault detection effectiveness of the proposed weighting based prioritized combinatorial testing.