On testing hypotheses in fuzzy surroundings

Abstract

Various statistical models have found their place in practical ap- plications; however, working with fuzzy-valued data still presents a challenge. This paper provides an overview of several well-known concepts of statistical hypothesis testing in fuzzy environments and introduces a new approach based on the comparison of triangular fuzzy numbers along the horizontal axis, i.e., on a horizontal fuzzy relation. The proposed method allows both the registered value of the test statistic and the reference critical value to be in fuzzy form and de nes fuzzy degree of acceptance of hypotheses.