Clustering by finding density peaks based on Chebyshev's inequality

To detect the cluster centers automatically by fast clustering using density peak detection and overcome the disadvantage of original algorithm: selecting the centers by visual identifying, Chebyshev's inequality is applied. The new clustering algorithm by finding density peaks based on Chebyshev's inequality (CDP) can get the judgment index by screening density and distance which are normalized. The judgment index can reflect the attributes of the data because cluster centers' judgment indices are much larger than others'. These points whose judgment indices are over the upper bound based on Chebyshev's inequality will be selected as the cluster centers. Then assign the remaining points by their nearest neighbor of higher density. The experimental results show its good performance for data sets in different shapes in comparison with other excellent clustering algorithms.