Improvement of K-means algorithm based on density
Lanlan Zhang, Jinshuai Qu, Minghu Gao, Meina Zhao · 2019
In view of the shortcomings of the traditional K-means method, the D-K-means algorithm is proposed in this paper. The algorithm adopts the concept of density number. The point set of high density number is extracted from the original data set as a new training set, and the point in the point set of the high density number is selected as the initial cluster center point. Then, using the method of geometric center points to update the cluster center points at high density points until convergence conditions are reached. Experiments show that the method can effectively avoid the local optimal situation of K-means clustering algorithm. On the other hand, the iterative number of iteration in the clustering process is reduced, and the stability and accuracy of clustering are improved.