A Novel Method for Visualization of Clustering Results
Chao Gao, Caiya Zhang · Communications in Statistics - Simulation and Computation · 2010
Sammon mapping is an approach of nonlinear dimension reduction and can be used for visualization. To avoid numerical complexity of the algorithm of traditional Sammon mapping, Kovacs and Abonyi (2004 Kovacs , A. , Abonyi , J. ( 2004 ). Visualization of fuzzy clustering results by modified Sammon mapping . Proceedings of the 3rd International Symposium of Hungarian Researchers on Computational Intelligence 177 – 188 . [Google Scholar]) proposed a modified Sammon mapping method. However, this improvement can only be applied to fuzzy clustering results. By using the property of Fermat point, we develop a new method in this article that can be applied to any clustering results. Different from other methods of visualization, we transfer information of clustering results into concentric circles around the Fermat points. So our procedure can demonstrate the data structure in a more informative way and the clustering results become easier to understand, especially for nonprofessionals. The effectiveness of the proposed method is studied by application to a real data in this article.