Clustering Centroid Selection using a K-means and Rapid Density Peak Search Fusion Algorithm

摘要

In the k-means algorithm, it is difficult to choose the $K$ value and the initial centroids of the clusters. To solve this problem, the K-CFSFDP method, which combines the “clustering by fast search and find of density peaks” (CFSFDP) algorithm and the k-means algorithm, was proposed. In this study, we obtained the optimal value of the hyperparameter $d_{c}$ by using the silhouette coefficient SIL and the error sum of squares $SSE$ to facilitate the selection of $d_{c}$ while testing the cluster centroid determined by the window selection method or the method that first sorts the products of $ho_{i}$ and $delta_{i}$ in descending order and then uses the slope change trend on the University of California-Irvine (UCI) dataset. We found that the window selection method was more stable and more effectively enhanced the clustering ability of the proposed k-means and CFSFDP fusion algorithm.