Hierarchical Clustering of Categorical Data using Improved Inter-Cluster Similarity

초록

This research proposed an improved hierarchical clustering algorithm for clustering categorical data. The traditional hierarchical algorithm was extended by calculating a similarity between clusters with a new inter-cluster similarity measure instead of the inter-individual measure used in the conventional algorithm. The proposed algorithm takes the cluster characteristics for categorical data into account in the similarity calculation, which is found to give better clustering results through experiments.

키워드