ABSTRACT

This chapter discusses clustering methods, the circumstances in which specific methods apply to global business intelligence, and the impact of these methods on global business intelligence. It describes various methods of clustering with respect to specific domains. Partition clustering relocates points from one partition to another. The advantage is that the quality of clustering can be improved with iterative optimization. The key challenges in stream clustering are the massive volume of online data and the continually evolving patterns of online data streams. Clustering has been applied to a wide variety of multimedia data such as image and video/audio. High-dimensionality brings in a special kind of challenge called the "curse of dimensionality" in which the data becomes increasingly sparse and presents various problems—such as, global optimization difficulty increases exponentially. Clustering validation makes an evaluation about how good the clustering results are. In data discovery for global business intelligence, clustering methods are powerful in finding unknown groups and hidden relationships.