PieceStack: Toward Better Understanding of Stacked Graphs

Area of Science:

• Data Visualization
• Human-Computer Interaction
• Information Visualization

Background:

• Stacked graphs are widely used for visualizing temporal sequences and their aggregations.
• Visual illusions in stacked graphs obscure the connection between individual layers and aggregated views.
• Existing methods fail to fully excavate the information contained within stacked graph formations.

Purpose of the Study:

• To introduce PieceStack, a novel visual analytic design for stacked graphs.
• To reveal the relevance of stacked graphs in understanding the intrinsic details of their displayed shapes.
• To address the limitations in interpreting the relationship between individual layers and aggregated data.

Main Methods:

• PieceStack interactively splits and reconstructs stacked graphs to interpret aggregation generation.
• A clustering algorithm partitions stacked graphs into sub-aggregated pieces based on layer trend similarities.
• Augmented encoding is used to visualize these pieces, aiding analyst exploration.

Main Results:

• The study demonstrates PieceStack's ability to clarify the formation of stacked graphs.
• Case studies and user studies validate the technique's effectiveness.
• Analysts can decompose and explore stacked graphs more effectively with PieceStack.

Conclusions:

• PieceStack offers a new approach to understanding the formation and details within stacked graphs.
• The technique mitigates visual illusion issues, enabling deeper data excavation.
• PieceStack enhances the analytical utility of stacked graph visualizations.