TitleHierarchical Spherical Clustering
Publication TypeJournal Article
Year of Publication2002
AuthorsTorra V, Miyamoto S
JournalInt. J. of Uncertainty Fuzziness and Knowledge-Based Systems

This work introduces an alternative representation for large dimensional data sets. Instead of using 2D or 3D representations, data is located on the surface of a sphere. Together with this representation, a hierarchical clustering algorithm is defined to analyse and extract the structure of the data. The algorithm builds a hierarchical structure (a dendrogram) in such a way that different cuts of the structure lead to different partitions of the surface of the sphere. This can be seen as a set of concentric spheres, each one being of different granularity. Also, to obtain an initial assignment of the data on the surface of the sphere, a method based on Sammon's mapping has been developed.