K-means Algorithm Based on Fitting Function

Abstract

Abstract. The K-means algorithm has the shortcomings of being sensitive to the initial clustering center, and in order to overcome this drawback, in this paper ,on the basis of the combination of data density and the optimal distance , a new definition of fitting function is made and then a kind of K-means algorithm based on fitting function is proposed. By utilizing the fitting function to select the initial clustering center, the selection of the initial cluster centers can be made as much close to the real sample clustering centers as possible. The experiments proved that, the K-means algorithm based on fitting function reduces the number of iterations and enhances the stability of the algorithm, as well as improves the efficiency of the algorithm.