Animating multidimensional scaling to visualize N-dimensional data sets

Abstract

Many techniques have been developed for visualizing multivariate (multidimensional) data. Most, if not all, are limited by the number of dimensions which can be effectively displayed. Multidimensional scaling (MDS) is an iterative non-linear technique for projecting n-D data down to a lower number of dimensions. This work presents extensions to MDS that enhance visualization of high-dimensional data sets. These extensions include animation, stochastic perturbation, and flow visualization techniques.