Quantitative Comparison of Treemap Techniques for Time-Dependent Hierarchies

Abstract

Hierarchical data in which leaf nodes have associated attributes appear in several applications. Rectangular Treemaps (RTs) were designed to display this type of data by tightly packing cells representing the tree nodes, with additional information stored in the cell sizes and/or colors. For dynamic hierarchies, RTs have to obey several requirements regarding the optimal aspect ratio of cells, but also the stability of layout in presence of data changes. While static RTs studied the first requirement well, far less information is available on how RTs behave vs both requirements for dynamic data. We study how four known RT methods compare over several real-world dynamic hierarchies, and highlight how recent RT methods can be adapted to effectively handle optimal cell ratios and stability.