A Declarative Rendering Model for Multiclass Density Maps.

Area of Science:

• Data Visualization
• Computer Graphics
• Information Science

Background:

• Traditional multiclass maps use shape or color for categorical attributes, which scales poorly with increasing data density.
• Overplotting in multiclass maps necessitates data aggregation, potentially obscuring underlying patterns.
• Existing aggregation methods in cartography often lack expressiveness and scalability for complex datasets.

Purpose of the Study:

• To introduce multiclass density maps as an improved visualization technique for categorical data.
• To present a declarative model for specifying and rendering multiclass density maps.
• To demonstrate the expressiveness, scalability, and dynamic reconfiguration capabilities of the proposed model.

Main Methods:

• Developed multiclass density maps by computing multiple 2D histograms for each category.
• Introduced a declarative JSON grammar with associated visual semantics for specifying multiclass density map visualizations.
• Implemented the model for efficient rendering in web browsers and dynamic reconfiguration for data exploration.

Main Results:

• Multiclass density maps effectively address overplotting and improve readability for large datasets.
• The declarative model supports a wide design space and can reproduce existing aggregated map examples.
• The model enables dynamic reconfiguration for interactive data exploration without raw data recomputation.

Conclusions:

• Multiclass density maps offer a scalable and expressive solution for visualizing multiclass data.
• The declarative model provides a flexible and efficient framework for implementing and exploring these visualizations.
• This approach enhances the capabilities of multiclass map visualization for handling large and complex datasets.