Estimating the correct number of clusters by cluster validity index in cluster analysis is highly susceptible to noise data
separation among clusters and clustering algorithm
so the correctness of the estimated number of clusters is difficult to be guaranteed. In order to overcome this problem
validity index is used to estimated number of clusters in original dataset and reduced dataset based on the data reducing method proposed in reference[1 ]
the result demonstrate the method can enhance separation among clusters and effectively determine the optimal number of clusters.
An extensive comparative study of cluster validity indices [J] . Olatz Arbelaitz,Ibai Gurrutxaga,Javier Muguerza,Jesús M. Pérez,I?igo Perona. Pattern Recognition . 2013 (1)
Data clustering: 50 years beyond K-means [J] . Anil K. Jain. Pattern Recognition Letters . 2009 (8)