Graph-Based Visual Analytic Tools for Parallel Coordinates

Parallel coordinates is a fundamental visualization technique in multivariate data visualization. Visual clutter is one of the inherent weaknesses in parallel coordinates. In this paper, we present two visual analytic tools, the Selection Graph and the Relation Graph, to reduce the visual clutter. The Selection Graph is a brushing tool which helps users highlight the regions of interest. The Relation Graph organizes clusters in a structural manner, providing an intuitive interface for users to explore relations among clusters. Both tools neither distort nor filter the underlying data in parallel coordinates. The experiments on several real datasets demonstrate the effectiveness of our tools.