Donut plots that is – the new hot plot that has displaced the old pie chart. What is it with us and fatty dessert terms for terrible plotting concepts?
Donuts are a complete waste of space and are seemingly ubiquitous in dashboards these days. They serve little purpose other than to dilute data density while splashing useless color across an interface. Their complete lack of utility should have them outright banned from from the tool box of any serious data scientist.
So here’s a graphic I encountered recently:
There are a few problems here:
- You have to roll over the graphic to see the actual numeric values.
- You cannot read or see the smallest values.
- If you are colorblind like me, you have a really hard time mapping the color to the legend.
- Speaking of the legend, your eye is bouncing from left to right then back to the left each time you try to understand what value you are looking at.
In short, this “pretty” and “interactive” graphic requires the viewer to do a lot of work just to get a basic understanding of what is going on with the data. It’s exhausting, a waste of time, and a waste of pixels.
So let’s look at what we can do with a variant of a good old fashioned table with a few enhancements:
Now you can see all the jelly that was jammed into that donut. The original plot’s legend couldn’t even present the smallest contributors or “sources” that were actually represented in there. Now in one simple table, without any user required activity, we can:
- See the critical data, sorted from highest contributing to lowest moving down the table,
- See the actual value of the record count contributed by the source,
- See the relative contribution represented as a percentage,
- See the relative contribution represented as a bar,
- and see ALL the data contained in the graphic.
Here they are side by side:
So give some consideration to abandoning pies and donuts in graphical representations of data; they belong on your plate, not in your dashboards or summary reports.
Long live the sorted table!