A model for structure-based comparison of many categories in small-multiple displays

Area of Science:

• Data Visualization
• Information Visualization
• Human-Computer Interaction

Background:

• Multivariate data, encompassing both categorical and numerical information, is common across many application domains.
• Small multiple displays facilitate data comparison through juxtaposition, but overlay or difference encoding requires reference specifications.

Purpose of the Study:

• To present a formal model for defining semantically meaningful comparisons within small multiple displays.
• To enable comparisons by overlay or explicit encoding of computed differences for multivariate data.

Main Methods:

• Developed a formal model based on pivotized data hierarchically partitioned by categories.
• Proposed two reference specification alternatives: absolute (fixed reference category) and relative (semantic ordering).
• Supported multi-level hierarchical comparisons, including aggregated summaries.

Main Results:

• The model allows for flexible and meaningful comparisons within small multiple displays.
• Demonstrated applicability across various visualizations for overlay and difference encoding.
• Enabled comparisons to fixed references or preceding/succeeding categories in a semantic order.

Conclusions:

• The proposed formal model enhances the analytical power of small multiple displays for multivariate data.
• Offers robust methods for comparing categories at multiple hierarchical levels.
• Facilitates deeper insights through structured and semantically grounded data comparisons.