9.2.1: Treemaps (Hierarchical Area Charts)

Treemaps (Hierarchical Area Charts)

A treemap is a space-filling visualization composed of nested rectangles that represent a hierarchical structure [7]. Each rectangle’s area is proportional to a quantitative value, and rectangles are arranged within parent rectangles to show part-to-whole relationships within a hierarchy. For example, a treemap could show a company’s revenue by region (top-level rectangles), and within each region rectangle, subdivisions by product category. Treemaps were originally developed by Ben Shneiderman as a way to visualize hierarchical data in a compact form, especially useful when screen space is limited. The strength of a treemap is its ability to display many categories simultaneously while indicating their relative size – one can quickly see which categories are largest or smallest at each level of the hierarchy.

When to use a treemap: Treemaps are ideal for hierarchical data where you want to show parts of a whole at multiple levels. They work best if there are a few major categories with potentially many subcategories, and you need to convey the proportion each contributes to the total. A classic use case is budget visualization: the overall budget is subdivided into departments (each a rectangle sized by department budget), and within each department, rectangles for individual projects. Treemaps can effectively show the largest budget items and how sub-items roll up. They have been used in contexts like electoral vote distributions (e.g. showing electoral votes by state in a treemap, with states nested in political party categories) and disk space usage (visualizing folders and files on a hard drive).

However, treemaps have limitations. They can become hard to interpret if there are too many tiny categories or if many values are of similar magnitude. In such cases, the rectangles may appear almost equal, and labels or color distinctions become crowded, leading to a cluttered chart. Thus, it’s recommended to use treemaps when there is a distinct hierarchy and a significant variance in the values so that the area differences are apparent. If precise comparisons of a single metric across many categories are needed, a bar chart might still be clearer (treemaps sacrifice some precision for compactness and hierarchical display). In summary, use treemaps for overview of large hierarchical distributions – they excel at showing the big picture of how a whole breaks down into parts (and sub-parts), especially when graphical real estate is at a premium.