Research and Improvement Method Based on k-mean Clustering Algorithm

Abstract

In view of the sensitivity of the traditional mean algorithm to outliers and noise points, an improved mean algorithm is proposed in this paper, which is based on the density of the distribution of objects in space. In the measurement of density, the sensitivity of clustering effect to initial parameters is reduced. The improved algorithm can filter the "noise" data and discover the clustering of arbitrary shapes, which is obviously superior to the standard mean algorithm.