On Cluster Validation for Detecting the Number of Clusters in a Data Set

Free registration required

Executive Summary

In this paper, the authors analyze different techniques for cluster validation to detect the number of clusters in a dataset and introduce a new combination approach to cluster validation. The proposed algorithm is a combination of several validation indexes, which are used to simultaneously evaluate different partitions of a dataset generated with different clustering techniques and object distances. The term "Cluster analysis" refers to the framework of algorithms used to capture the underlying group structure of unlabelled data sets.

  • Format: PDF
  • Size: 216.3 KB