Visual Abstraction and Exploration of Multi-class Scatterplots

Area of Science:

• Computer Science
• Data Visualization
• Human-Computer Interaction

Background:

• Scatterplots are essential for visualizing datasets, identifying patterns like outliers, clusters, and correlations.
• Visualizing multi-class scatterplots often leads to overdraw, hindering efficient data analysis and interpretation.

Purpose of the Study:

• To develop a novel visual abstraction scheme for efficient multi-class scatterplot visualization.
• To address the challenge of overdraw in scatterplots with numerous data classes.
• To enhance the density contrast and preserve key features of the dataset.

Main Methods:

• A hierarchical multi-class sampling technique is employed for feature-preserving simplification.
• Color optimization is performed considering multi-class point distributions to improve density contrast.
• A visual exploration system is designed for interactive inspection and quantitative analysis.

Main Results:

• The proposed method effectively simplifies complex multi-class scatterplots, reducing overdraw.
• Optimized color schemes enhance the visibility of different data classes and their distributions.
• The system facilitates both visual inspection and quantitative analysis of complex datasets.

Conclusions:

• The developed visual abstraction scheme significantly improves the efficiency of analyzing multi-class scatterplot data.
• The approach offers a practical solution for handling overdraw issues in dense, multi-class visualizations.
• The system demonstrates effectiveness across various challenging datasets, validating its utility.