Cluster Validation of Evolutionary Clustering Algorithm for Multivariate Datasets

Abstract

Clustering is an unsupervised technique that is used when the information about data is inadequate. Clustering algorithms identify clusters in the dataset and acquire information on the number of clusters to be identified as a prior parameter. In the present work, an adaptive evolutionary clustering algorithm (AECRAM), which evolves automatically, was tested to identify the number of clusters in a multivariate dataset. It is important to test whether the identified clusters are actual clusters or not, and if difference exists in them, then to find out these differences, relative to their actual clusters using cluster validity indexes. Using adaptive evolutionary algorithm cluster validation was done in this work and compared with K-means and EKM algorithms for their relative efficiency. The observations made in the study clearly indicated that the proposed adaptive clustering algorithm could detect correct and accurate number of clusters in the multivariate datasets. The quality of the clusters identified by the AECRAM has been found to be superior to the other two algorithms used. The present study is also useful in other research where relative cluster validations among adaptive evolutionary clustering algorithms are required.

Keywords: Cluster, Validity Index, Dunn Index, Malignant, Benign

View PDF