Visualization of a set of parameters characterized by their correlation matrix

An approach to visualization of a set of parameters characterized by their correlation matrix has been proposed. It integrates two methods for data mapping: Sammon's mapping and self-organizing map (SOM). They are based on different principles, and, therefore, supplement each other when they are used jointly. It is shown experimentally that some (sometimes sufficient) knowledge on a set of parameters may be obtained by using individual methods. However, in most cases the necessity and quality of their joint use is unquestionable - this allows us to observe the same data set from various standpoints and to extend our knowledge on the object of investigation. (C) 2001 Elsevier Science B.V. All rights reserved.