In the past two weeks, we showed you some of the ways designers get creative with pie charts and bar charts. Another class of visualizations that designers use is based on area representations of data. These charts include everything from scaled bubbles to ISOTYPE visualizations, and they vary widely in their effectiveness. Let’s take a look at seven ways designers use area to display data, and dissect if and how they are harming the communication of the data.
- Area Bubbles are the simplest of these systems for encoding quantity to area. These are not super effective ways to accurately communicate quantity, but could be used reasonably well for a handful of values, especially when sorted by quantity.
- Area Icons are a little more complex. They have the advantage of simultaneously labeling and showing data they represent, but the complexity of the icons can interfere with the effectiveness of encoding data to area.
- Fractional Icons show quantities in a part to whole relationship, similar to icon pie charts, but they are often segmented, much like a stacked bar. The technique is often executed incorrectly, failing to account for the odd shapes, and instead encoding data based on the height or width of each segment.
- ISOTYPE Visualizations are another common way of showing counts. They discretize the count, and often a single icon will represent a higher quantity of the units (1 stick figure represents 10 people). The approach is good in many cases, but often has problems when comparing two icon sets with different ink densities. They are most effective when also used as bar charts encoding quantity to height or width instead of just area.
- Treemaps use area to encode quantity, and while they aren’t particularly useful for displaying hierarchies in a static visualization, they are good at showing different quantities in a flat hierarchy. They are great when used with reasonably proportioned rectangles, however applying them to icons, or using circles damages their effectiveness.
- Overlapped Areas are an interesting case. They damage the raw area encoding by overlapping, but if an area is contained entirely within another, they can still effectively convey the quantity they represent because the area can still be easily inferred.
- Euler Diagrams also encode quantity as area, however the overlapping sections in them actually serve a dual (sometimes triple or more) purpose. In addition to adding to the area of the parent regions, the sub-region also represents a quantity of the subset of each.
These are all different approaches for using area to show data, and many of them also show different relationships within the data. They all use area as the primary data encoding, and their effectiveness at communicating that data varies from technique to technique. Choose your area-based visualization wisely, and be careful to calculate the areas correctly. Drew Skau is Visualization Architect at Visual.ly and a PhD Computer Science Visualization student at UNCC with an undergraduate degree in Architecture. You can follow him on twitter @SeeingStructure