Title: Interactive Hierarchical Displays: A General Framework for Visualization and Exploration of Large Multivariate Data Sets

Authors: Jing Yang, Matthew O. Ward, and Elke A. Rundensteiner

Abstract: Numerous multivariate visualization techniques and systems have been developed in the past three decades to visually analyze and explore multivariate data being produced daily in application areas ranging from stock markets to the earth and space spaces. However, traditional multivariate visualization techniques typically do not scale well to large multivariate data sets, with the latter becoming more and more common nowadays. This paper proposes a general framework for interactive hierarchical displays (IHDs) to tackle the clutter problem faced by traditional multivariate visualization techniques when analyzing large data sets. The underlying principle of this framework is to develop a multi-resolution view of the data via hierarchical clustering, and to use hierarchical variations of traditional multivariate visualization techniques to convey aggregation information about the resulting clusters. Users can then explore their desired focus region at di erent levels of detail, using our suite of navigation and ltering tools. We describe this IHD framework and its full implementation on four traditional multivariate visualization techniques, namely, parallel coordinates [9,19], star glyphs [16], scatterplot matrices [4], and dimensional stacking [12], as implemented in the XmdvTool system [18,13,6,7]. We also describe an empirical evaluation that veri ed the e ectiveness of the interactive hierarchical displays.

Key words: Large-scale multivariate data visualization, exploratory data analysis, hierarchical data exploration

Source: Computers and Graphics Journal, 2002.

Download: PDF document