Fuzzy p-mode prototypes: A generalization of frequency-based cluster prototypes for clustering categorical objects

Frequency-based cluster prototypes were developed in to cluster categorical objects, based on the simple matching dissimilarity measure. This paper describes a generalization of the frequency-based prototypes in the same framework of the fuzzy C-means clustering algorithm for the objects of mixed features. In the general fuzzy C-means clustering algorithm, a cluster prototype, called fuzzy p-mode prototype, at the categorical feature level is expressed as a list of p labels that have larger frequencies than others. The convergence of the general fuzzy C-means clustering algorithm under the optimization framework is proved. It is also explained through experiments over real object sets that sizes of fuzzy p-mode prototypes and fuzzification coefficients affect clustering performance.