Guidelines
Suggest edit

Assign categorical symbols so perceptual differences match group differences

For cluster-oriented reading of categorical group relations, use distance-preserving shape or color assignments on categorical encodings to improve fidelity and mitigate arbitrary similarity cues for viewers interpreting how groups relate.

  • purpose:refine
  • basis:empirical
  • task:cluster
  • data:categorical
  • quality:fidelity:use
  • lever:encoding
  • operator:association

advice

Distance-preserving assignment

Assign category colors or shapes so perceptual distances mirror the distances between the groups they encode. For example, when group-to-group connection strengths or other pairwise distances are known, give closely related groups more similar symbols and distant groups more distinct ones instead of mapping categories to the palette arbitrarily.

reason

Why matched assignments help

Distance-preserving assignments turn visual similarity into a usable cue about data similarity. Viewers can then read group relations from the encoded symbols, not just group identity.

Mechanism: When perceptual similarity tracks group similarity, the encoding preserves more of the underlying structure and makes relational patterns easier to interpret.

Evidence: The paper presents visual embedding with perceptual kernels and shows color and shape assignments chosen to reflect distances among community clusters, so visually similar symbols represent more closely related groups. A later review collates this paper as empirical evidence on nominal shape, color-hue, and area encodings for cluster-oriented visualization design (Zeng & Battle, 2023; Demiralp et al., 2014).

context

Use when group relations matter

  • User Goal: Show not just which group an item belongs to, but also how groups relate.
  • Task: Read relative closeness or separation between groups from categorical encodings.
  • Data: Categories or clusters with a known pairwise distance, similarity, or connection-strength structure.
  • Chart Setting: Color or shape already encodes group membership.
  • Audience: Viewers need to interpret relations among groups from the same display.
  • Success Criterion: Visually similar symbols correspond to closer groups, and visually distinct symbols correspond to farther groups.

exceptions

Do not use when there is no group-distance structure to preserve

Break it when: The categories do not have a meaningful pairwise similarity or distance structure. Why: The assignment rule depends on preserving an existing structure; without one, there is nothing for perceptual similarity to represent.

costs

Costs of matched assignment

Sacrifice: You need both a group-distance model and a perceptual kernel. Risk: Preserving one chosen structure can leave other possible structures unrepresented. Mitigation: Decide which group relation is most important before optimizing the assignments.

mistakes

Common assignment mistakes

Mistake: Map categories to colors or shapes in arbitrary order when the groups have known similarities or distances. Why it fails: Visual similarity then carries accidental meaning instead of reflecting the intended group structure.

check

Check whether the mapping preserves structure

Failure Sign: Two close groups get very different symbols, or two distant groups get very similar ones. Quick Check: Compare a few nearest and farthest group pairs against their assigned colors or shapes. Stronger Test: Compare the group-distance matrix to the perceptual distances induced by the final assignments and check whether the ordering broadly agrees.

fix

Fix the category mapping

  • Derive the pairwise distance or similarity matrix for the groups you want to relate.
  • Choose the perceptual kernel for the encoding channel you will use.
  • Optimize the category-to-symbol assignment against those two distance structures instead of using arbitrary category order.
  • Revise the legend mapping so the final color or shape assignments preserve the intended group relations.

References

Demiralp, Ç., Bernstein, M. S., & Heer, J. (2014). Learning Perceptual Kernels for Visualization Design. IEEE Transactions on Visualization and Computer Graphics, 20(12), 1933–1942. https://doi.org/10.1109/TVCG.2014.2346978
Zeng, Z., & Battle, L. (2023). A Review and Collation of Graphical Perception Knowledge for Visualization Recommendation. Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems, 1–16. https://doi.org/10.1145/3544548.3581349