An automatic clustering algorithm based on DB criterion

Inspired by a recent algorithm on clustering, we proposed an improved algorithm which combines the Davies-Bouldin criterion to obtain the right number of the cluster centers automatically and output the right clusters. Davies-Bouldin criterion can describe the intra-class scatter and inter-class deviation value of the clustering result. In our algorithm, we firstly calculate the density and the distance of the sample points, which contain the information of the density distribution leading to the right clusters; Then, we choose two thresholds of the density and the distance to obtain the maximum number of the cluster centers; Finally, our algorithm automatically search the right number of the cluster centers through calculating the Davies-Bouldin value of every clustering result and choose the one which has the minimum Davies-Bouldin value. Experiments show that our algorithm can not only output the right clustering result when the sample points are disturbed and with special density distribution, but can also obtain the right number of the cluster centers automatically.