Hierarchical Edge Bundling
A radial network layout that uses a hierarchical grouping of nodes to bundle edges together — revealing high-level connection patterns between clusters while dramatically reducing visual clutter.
// 01 — The chart
What it looks like
A hierarchical edge bundling diagram showing software module dependencies. Highlighted edges reveal the strongest coupling between auth, api, db, and cache modules.
// 02 — Definition
What is hierarchical edge bundling?
Hierarchical edge bundling is a visualization technique for networks (graphs) where nodes are organized in a hierarchy — such as files in a directory tree, modules in a software system, or departments in an organization. Instead of drawing straight lines between connected nodes, edges are routed through the hierarchy and bundled together where they share common ancestors.
The result is a radial layout where leaf nodes sit around the circumference of a circle, and connections between them curve inward, following the branch structure of the hierarchy. When many edges share a similar path through the hierarchy, they naturally merge into thick bundles — making it easy to see which groups of nodes have strong inter-connections.
The key insight is this: bundling trades individual edge readability for a high-level overview of connection patterns. You sacrifice the ability to trace a single specific connection in exchange for seeing the overall structure of dependencies between groups — a worthwhile trade when the network has hundreds or thousands of edges.
Origin: Hierarchical edge bundling was introduced by Danny Holten in 2006 in his paper “Hierarchical Edge Bundles: Visualization of Adjacency Relations in Hierarchical Data.” The technique combines existing ideas from radial tree layouts and spline-based edge routing into a powerful new approach for dense network visualization.
// 03 — Anatomy
Parts of a hierarchical edge bundling diagram
// 04 — Usage
When to use it — and when not to
- Your nodes have a natural hierarchy (files in folders, packages in modules, teams in departments)
- The network has hundreds or thousands of edges and a node-link diagram would be unreadable
- You want to reveal inter-group dependency patterns at a high level
- Identifying strongly coupled clusters is more important than tracing individual edges
- You need an aesthetically appealing overview for presentations or publications
- Exploring software dependency structures or call graphs
- Your data has no natural hierarchy — the technique requires a tree structure to route edges
- You need to trace specific individual connections precisely — bundles obscure exact paths
- Your network is small (under 20 nodes) — a simple node-link diagram will be clearer
- Edge weights or directions are crucial to communicate — bundling hides these details
- You need users to discover the exact endpoints of each connection interactively
- The hierarchy is very flat (only one level) — bundling needs multiple levels to be effective
// 05 — Reading guide
How to read a hierarchical edge bundling diagram
Follow these steps to interpret a hierarchical edge bundling visualization.
Identify the hierarchy and groups
Look at the arrangement of nodes around the circle. Nodes that are near each other belong to the same group in the hierarchy. Labels or color may indicate these groups — such as modules, departments, or directories.
Find the thickest bundles
The thickest clusters of curves represent the strongest connection patterns between groups. These are your main insight — they show where the most inter-group dependencies or relationships exist.
Note which groups are heavily connected
Follow a thick bundle from one side of the circle to the other. The groups at each end have high cross-coupling. If two groups on opposite sides are connected by a dense bundle, they have a strong relationship despite being separated in the hierarchy.
Look for isolated groups
Groups with few or no edges crossing to other parts of the circle are relatively independent. In a software context, these are well-encapsulated modules. In an organizational context, these are siloed teams.
Adjust the bundling tension mentally
Consider that high bundling tension produces tighter, more abstract curves. If the visualization looks too cluttered, the tension may be too low. If it looks too abstract, individual connections may be hard to distinguish. Interactive tools let you adjust this in real time.
// 06 — Common mistakes
Mistakes to watch out for
Using it without a meaningful hierarchy
Hierarchical edge bundling requires a genuine tree structure to route edges through. If you assign an arbitrary or artificial hierarchy, the bundling patterns will be meaningless — edges will group together for no semantic reason, producing misleading visual clusters.
Setting bundle tension too high
Very high tension causes all edges to converge toward the center, making distinct bundles merge into an indistinguishable mass. You lose the ability to see which groups connect to which. Use moderate tension (typically 0.85) that balances readability with bundling benefit.
Ignoring node ordering around the circle
The order of leaf nodes on the circumference matters enormously. Placing related nodes apart forces their edges to cross the entire diagram unnecessarily. Always order nodes so that related items (siblings in the hierarchy) are adjacent.
Overloading with too many edge colors
Using many colors for different edge types creates a confusing rainbow. Hierarchical edge bundling is already visually dense. Limit color encoding to one or two dimensions (e.g., direction or weight) and use opacity to convey density instead of color.
Not providing interactivity for large graphs
Static hierarchical edge bundling diagrams with 100+ nodes are nearly impossible to read without interaction. Users need hover-to-highlight behavior (showing only edges connected to the hovered node) to make the diagram useful. Without this, the visualization becomes decoration rather than analysis.
// 07 — Real-world examples
Where you’ll see hierarchical edge bundling used
Software engineering: Codebase dependency analysis
Development teams use hierarchical edge bundling to visualize import dependencies between packages and modules. Thick bundles between two packages indicate tight coupling that may signal a need for refactoring. Tools like D3.js make these diagrams interactive, letting engineers hover over a module to see all its connections.
Software EngineeringBiology: Gene regulatory networks
Researchers studying gene expression organize genes by biological function (the hierarchy) and visualize regulatory relationships as bundled edges. This reveals which functional groups of genes interact most, providing clues about co-regulation and pathway dependencies.
BioinformaticsOrganizational network analysis
Large companies map communication patterns (emails, meetings, project collaborations) between departments using hierarchical edge bundling. The organizational chart provides the hierarchy, and edges represent cross-department interactions. Dense bundles reveal strong working relationships; sparse connections highlight organizational silos.
Business Analytics// 08 — At a glance
Quick reference
// 09 — Variations
Types of edge bundling
Hierarchical edge bundling is one of several edge bundling strategies, each with different strengths.
Radial bundling (standard)
The classic form — nodes arranged on a circle with edges bundled through the hierarchy center. Best for exploring symmetric relationships.
Force-directed edge bundling
Bundles edges using attraction forces rather than hierarchy. Works on any graph — no hierarchy required — but can be computationally expensive.
Directional bundling
Adds direction to bundled edges using color gradients or tapering, showing flow from source to target groups.
Multi-level bundling
Uses multiple hierarchy levels to create nested bundling, allowing zoom-in exploration of edge patterns at different granularities.
// 10 — FAQs
Frequently asked questions
What is a hierarchical edge bundling?+
Hierarchical edge bundling is a visualization technique for networks (graphs) where nodes are organized in a hierarchy — such as files in a directory tree, modules in a software system, or departments in an organization. Instead of drawing straight lines between connected nodes, edges are routed through the hierarchy and bundled together where they share common ancestors.
When should you use a hierarchical edge bundling?+
Use a hierarchical edge bundling when your nodes have a natural hierarchy (files in folders, packages in modules, teams in departments). It also works well when the network has hundreds or thousands of edges and a node-link diagram would be unreadable, and when you want to reveal inter-group dependency patterns at a high level.
When should you avoid a hierarchical edge bundling?+
Avoid a hierarchical edge bundling when your data has no natural hierarchy — the technique requires a tree structure to route edges. It is also a poor fit when you need to trace specific individual connections precisely — bundles obscure exact paths, or when your network is small (under 20 nodes) — a simple node-link diagram will be clearer.
Is a hierarchical edge bundling suitable for dashboards?+
Yes — a hierarchical edge bundling can work well in dashboards as long as the panel is large enough for readers to perceive the encoded values, has a clear title, and includes the legend or axis labels needed to interpret it.
What category of chart is a hierarchical edge bundling?+
Hierarchical Edge Bundling belongs to the Network family of charts. Charts in that family are designed to answer the same kind of question, so they often work as alternatives when one doesn't quite fit your data.
How do you read a hierarchical edge bundling?+
Start with the axis labels and legend, then look at the overall shape before zooming into individual marks. Compare prominent features against the rest of the data, and verify any conclusion against the underlying numbers when precision matters.