SIPTA Seminar by Beatriz Sinova: When robust statistics meets imprecise-valued data

ABSTRACT: Fuzzy-valued data are particularly useful for capturing intrinsic imprecision. However, most of the methods from the literature for their statistical analysis often fail in the presence of outliers, meaning that conclusions may be compromised under data contamination. Considering that contamination frequently appears in real-life experiments, several robust approaches for summarizing the central tendency and scale of fuzzy-valued data have already been proposed. Robust statistics has inspired these ideas, but also provided the mathematical framework to guarantee the robustness of the new measures. In this talk, the foundations of both robust statistics and the statistical analysis of fuzzy-valued data will be reviewed, and the focus will be set on showing the adaptation of classical robust statistics to imprecise-valued data. This talk is part of a series of seminars on imprecise probabilities that are organized by SIPTA, the "Society for Imprecise Probabilities: Theories and Applications". We also organize conferences and schools, provide documentation and maintain a mailing list and blog. More information is available at http://sipta.org. Info on the SIPTA seminars in particular is available at http://sipta.org/events/sipta-seminars Contents 00:00 - Start 02:07 - Introduction to fuzzy-valued data 20:03 - Robust statistics 25:48 - Robust measures for fuzzy-valued data 44:56- Simulations 50:09 - Future directions