A composite multidimensional visualization method based on eCLPCs

In this paper, we propose a composite multidimensional visualization method based on an enhanced circular layout parallel coordinates (eCLPCs) design, integrated with points and lines, by taking advantage of parallel coordinates, Andrews Curves, scatterplots and radial coordinates. Following the design of circular layout parallel coordinates, we give a scheme including three layers, from outside to inside separately as Andrews Curves connecting with parallel coordinates to visualize the linear dependence of several selected dimensions, line connections or points congregations to illustrate the relations of a chosen dimension with all other dimensions, and radial coordinates realized with Spring Model to enhance the overall clustering effects. Also, a series of interactions have been developed in our work including keyboard control, mode transformation, filtering and searching operations, etc. Experiments on a classic car dataset and a grade report dataset prove that our scheme is efficient to solve multidimensional problems from multiple aspects.