Dynamic fuzzy clustering using fuzzy cluster loading

Abstract

When we obtain clusters through the classification of a given data it is important to interpret the meaning of the obtained clusters. This is particularly true in the clustering of 3-way asymmetric similarity data. This is true because the asymmetric property and the structure of similarity in each cluster are changed over the time periods (or situations), and the interpretation of the obtained clusters is also changed according to the time periods (or situations) or according to the direction of asymmetry.

Therefore, this paper proposes a model for capturing the dynamic interpretation of the obtained clusters according to the time periods (or situations) or the direction of asymmetry of 3-way asymmetric similarity data using the idea of the dynamic fuzzy clustering model (M. Sato-Ilic, Y. Sato, “A dynamic additive fuzzy clustering model”, in Advances in Data Science and Classification, A. Rizzi et al., Eds, Berlin: Springer-Verlag, 1998) and fuzzy cluster loading (M. Sato-Ilic, “On kernel based fuzzy cluster loadings with the interpretation of the fuzzy clustering result”, Int. J. Comput. Numer. Anal. Appl., 4(3), pp. 265–278, 2003).

Keywords:

• Asymmetric similarity data
• Cluster loading
• Dynamic change
• 3-way data
• Asymmetric aggregation operator
• Super-matrix