Performance Analysis of Various Entropy Measures in Categorical Data Clustering

Categorical data clustering is gaining importance in recent years. Shannon entropy based similarity and dissimilarity measures are popular for clustering categorical data. However, there are several other generalized entropy measures like Renyi and Tsallis. The paper presents a thorough comparative study of performance of various entropy measures in clustering categorical data. Exhaustive experimental analysis reveals that Renyi entropy measure outperforms other measures in clustering categorical data.