Research On An Improved K-means Algorithm

The paper introduces and summarizes the typical clustering algorithms which are suitable for large data,discusses their main advantages and disadvantages,and proposes a feasible improvement method based on them.The main methods using k-nearest neighbor algorithm,based on the common data point density,defines the data point density of new,this definition is the key steps of the new algorithm,the data can be set according to the density of decommitment.In addition,the definition of standard radius,used to generate the adjacency matrix,the initial point selected to meet the conditions and will not produce empty clusters.Finally,in order to explain the improvement results,a two-dimensional data set is selected for testing.