A Study on Edge Bundling Technology in Parallel Coordinates

Visual clutter caused by a large data set seriously hampers the presentation accuracy of information in parallel coordinates. The number of items in every cluster cannot be accurately expressed in the overlap region in previous methods. In the paper, we propose an edge-bundling method to solve the problem. Each cluster in the axis is first divided to several groups in terms of the connection. Then, the connection between groups is used to reorder groups. Finally, all groups are rendered independently with different colors. In order to enhance visual effects, we design an interactive interface to highlight the selected group. Experiments show that our approach is effective to visualize a large data set in parallel coordinates and the visual continuity of the connection relation is preserved well.