B1026
Title: On the robustness of fuzzy-valued M-estimators
Authors: Beatriz Sinova - University of Oviedo (Spain) [presenting]
Stefan Van Aelst - University of Leuven (Belgium)
Abstract: Fuzzy-valued data have become a suitable tool to deal with imprecision in real-life experiments. During the last years many statistical techniques and procedures have been adapted to cover this new situation, such as regression analysis, clustering or hypothesis testing. Most of these developments are based on the Aumann-type mean as measure of location, but this fact leads to non-robust statistical conclusions due to the high sensitivity of the mean, inherited from the real settings. To overcome this important drawback, the successful and classical approach of M-estimators of location will be extended to deal with fuzzy-valued data and its main features in terms of robustness will be analyzed.
