A Unified Metric for Categorical and Numerical Attributes in Data Clustering

Most of the existing clustering approaches concentrate on purely numerical or categorical data only, but not the both. In general, it is a nontrivial task to perform clustering on mixed data composed of numerical and categorical attributes because there exists an awkward gap between the similarity metrics for categorical and numerical data. This paper therefore presents a unified metric for data clustering, in which the attributes are in either one of the three types: numerical, categorical, and their both. We firstly present a general clustering framework based on the concept of object-cluster similarity. Then, a unified metric of object-cluster similarity is presented. Finally, an iterative clustering algorithm is developed, which is directly applicable to the three data types stated above without any adjustment. Experimental results show the efficacy of the proposed approach. A Unified Metric for Categorical and Numerical Attributes in Data Clustering Yiu-ming Cheung Department of Computer Science Hong Kong Baptist University Hong Kong SAR, China [email protected] Hong Jia Department of Computer Science Hong Kong Baptist University Hong Kong SAR, China [email protected]