A new metric on parallel coordinates and its application for high-dimensional data visualization

Abstract

High-dimensional data visualization is a changing task with many applications in a various fields of sciences. Parallel coordinates is one of the most widely used information visualization technique for multivariate data analysis and high-dimensional geometry. The dimension ordering is an original problem for exploring structures in a high-dimensional data space. In this paper, we propose a new metric for measuring distance between two line-segment on the parallel coordinates. The metric is suitable and effective on the parallel coordinates. We use our metric distance for finding an optimal dimension ordering on the parallel coordinates. Finally, we demonstrate our method can be applied to visualize clusters in high-dimensional data on the parallel coordinates.