Введение в область нечеткой кластеризации качественных и количественных данных

Abstract

In many applications of clustering objects described by both quantitative and qualitative characteristics. A variety of algorithms have been proposed for "uncontrolled classification", which is the best thing for fuzzy partitions, and described the cluster prototypes. However, most of these methods are designed for data sets that can be represented in a limited type (only classes or only the metric). We propose a new approach to fuzzy clustering based on the probability measure distances.