Agrupamento não supervisionado fuzzy C- Means

Abstract

The clustering task aims to identify and approximate similar data. A cluster is a collection of data similar to each other, but different from the other records in the other clusters. In this context, many of these clusters have already been proposed and are still the origin of several scientific studies aimed at obtaining the best data separation. Among the best known methods are k-means and fuzzy c-means. Both have deployments in various commercial and free packages. However, ordinary users tend to use such methods indiscriminately, without knowing their implications and differences. This work aims to contribute precisely in this sense, to give clarity and practical examples of the behavior of the method fuzzy c-means, which is based on the theory of fuzzy sets.