Data Representativeness Based on Fuzzy Set Theory

Frédéric Blanchard, Philippe Vautrot, Herman Akdag et Michel Herbin

This paper presents an original de nition of data representativeness.The representativeness of each datum in a dataset is a meaningful notion quanti ed by a degree computed by aggregating fuzzy subsets. These fuzzy subsets are obtained by fuzzifying data in a robust way. We illustrate the usefulness of the representativeness by presenting applications for statistical location estimation,and for cluster analysis.

Mots clés

Data Representativeness, Fuzzy Set Theory, Robustness, Data Analysis, Clustering, Multidimensional data, Multivariate Analysis, Ordinal Analysis, Rank Transformation.