An improved clustering algorithm based on local density

Abstract

Clustering analysis is an important and challenging research field in machine learning and data mining.A fast and effective clustering algorithm based on the idea of local density was proposed by Alex.But the number of clusters and cluster centers in the algorithm were determined by hand.Therefore,the candidates of cluster centers based on local density were firstly selected and then density connectivity method was used to optimize the candidates.The classes of samples are the same as the nearest center with bigger local density.Experiments show that the proposed method has a better cluster efficiency and can handle the problems of uncertain cluster number and cluster centers.

Keywords

• local density
• cluster center
• evaluation criterion