The Study On The Clustering Algorithms

With the rapid development of the Internet, people begin to access more and more information. But it is difficult for people to understand the patterns of large data. In order to obtain the patterns of large data, people introduce the clustering technologies. The paper summarizes and analyzes clustering technologies. First, the paper introduces main clustering algorithms systematically and analyzes the key techniques of clustering. Second, the paper gives two improved algorithms compared with the original ones. One is an improved algorithm of K-means which gives a new method of selecting initial centroids and the improved algorithm improves the clustering accuracy, the other is an improved algorithm of DBSCAN which gives an automatic process of determining parameters. Through the analysis of data and experimental results, we find that the improved algorithms are more suitable for the distribution of the data and extend the ability of the original algorithms.