A Bi-Robust Test for Vague Data

Przemys law Grzegorzewski.

Nonparametric statistical tests are robust to disturbances of the assumptions on the underlying distributions. However, in the presence of vague data, modelled by fuzzy sets, the problem of robustness is much more complicated. Here not only distribution-free methods but tests which are also not sensitive to the particular choice of the membership functions applied for modelling vague concepts would be desirable. A construction of such bi-robust test is suggested in the paper.

PDF full paper