Yan Holtz logo ← All issues
Dataviz Universe

Hierarchical edge bundling

Today I suggest we take a quick look at this kind of graph. Do you know what it is?

Visualizing information flow in science by Moritz Stefaner

It's called hierarchical edge bundling, and it might just be my favorite graph type—it's visually stunning!

Let's explain what it is in three minutes or less.

1. Hierarchy

Think of a folder on your computer with subfolders and files inside each.

This descending structure is a hierarchy, where each item is a node, and the smallest items are leaves.

We can represent this as a circular dendrogram, displaying all files arranged around a circle:

A hierarchy visualised as a circular dendrogram

2. Connections

Now, imagine File 1 calls File 9 to run.

This creates a connection between them. Let's represent it with a line!

A straight line showing a connection between two leaves

Drawing this with straight lines would get messy fast with multiple connections.

Luckily, there's a better solution!

3. Bundling

Instead of straight lines, hierarchical edge bundling curves the connection to follow the hierarchy between two elements. Like this!

Curves following the hierarchy declutter the figure

Connections within the same group bundle together, reducing clutter and creating a more meaningful—and beautiful—visual.

As the inventor of this technique put it, it's like:

...bundling your electrical wires together to reduce clutter, while fanning them out at their terminus to plug them in.
Result with many connections

Wrapping Up

I hope you've added hierarchical edge bundling to your mental toolkit for future dataviz projects!

If you're interested in more explanations like this, check out my data-to-viz.com project, where I cover a range of visualization types. I also have tutorials in R, Python, and D3 to help you create these graphs from scratch.

P.S. Have you tried hierarchical edge bundling before? I'd love to see examples to feature on dataviz-inspiration.com! Feel free to drop me an email.


See you next week!
Yan

Enjoyed this?

Get one short, actionable dataviz tip every Saturday. Join 19,000+ data people.