Enhancing parallel coordinate plots

Abstract

Effective visual representation of data sets with many continuous variables is not easy even with modern statistical graphic tools. Among them, the parallel coordinate plot (PCP) is especially popular these days due to its conceptual simplicity and compact appearance. However, the effectiveness of PCPs is limited because of its dependency on the order of variables and its peaked up-and-down patterns. Recently, the order dependency problem has been solved by Hurley [Hurley, C. B. (2004). Clustering visualizations of multidimensional data. Journal of Computational and Graphical Statistics, 13, 788-806]. In this paper, we will propose an enhanced version of PCP (ePCP) for an optimally re-ordered list of variables that has proportionate spacings between variables and data points connected by "near smooth" curves. © 2008 Elsevier Ltd. All rights reserved.