The improved research on k-means clustering algorithm in initial values

This paper deeply works over the aspect that the k-means clustering algorithm is very sensitive to the initial values. In order to improve the dependence on the initial values, it proposes a new algorithm called K-means clustering algorithm based on iterative density (hereinafter referred to as IDKM). Through continuous modification to density threshold, it gets the more clustering centers, and merges them until the specified number of clustering center is met. IDKM algorithm is applied to the IRIS data set for clustering analysis, and then the result proves that the improved algorithm optimizes the dependence; Finally, IDKM is applied to Student achievement data set, the analysis of the clustering results guides students to study, it realizes the application of K-means clustering algorithm on data mining.