Large-Scale Data Visualization with Two-Variate Level-of-Detail Control

Space-filling approach, such as a series of TreeMaps algorithms, is one of the popular approaches in hierarchical data visualization. The original TreeMaps [1] represents the hierarchy as nested belt chart (so called slice-and-dice), and then it has been improved by squarifying or ordering the subregions representing the nodes of a tree. These algorithms represent quantitative properties of nodes as areas of subregions; however, it is often required that leaf-nodes of a tree is displayed as equally-sized icons. Quantum TreeMaps [2] is a variation of TreeMaps that represents leaf-nodes as equallysized icons or thumbnails while it attempts to quantize the widths and heights of rectangular subregions so that the icons are wellpacked inside them. Against these series of TreeMaps divides display spaces by top-down algorithms, another space-filling technique [3] represents trees as nested rectangular regions and equallysized icons applying a bottom-up algorithm. We are applying space-filling hierarchical data visualization techniques to the large-scale table data, as shown in Figure 2, which is supposed as following.