Abstract:

To deal with the deficiencies of a clustering algorithm for fast finding and searching of density peaks, an automatically and fast finding of clustering centers clustering algorithm is proposed, which adopts linear regression and residual analysis and optimizes sample density values. The algorithm uses sample’s nearest neighbors information to measure point density for improving clustering centers stability, and it uses linear regression and residual analysis to choose clustering centers fast and automatically and removes subjectivity of artificial selection. Theoretical analysis and contrast experiments show that the proposed algorithm can overcome deficiencies of the original algorithm, and the results of clustering and calculation time is better than the original algorithm, the density based spatial clustering of applications with noise (DBSCAN) and the K-means algorithm.