Dashboard Vision: Using Eye-Tracking to Understand and Predict Dashboard Viewing Behaviors

Area of Science:

• Human-Computer Interaction
• Information Visualization
• Cognitive Science

Background:

• Dashboards are crucial for data visualization, but design's impact on user engagement is understudied.
• Existing saliency models are limited to single visualizations, not complex dashboards.
• Designers often rely on intuition due to a lack of empirical data on dashboard interaction.

Purpose of the Study:

• To quantify user viewing patterns on dashboards using eye-tracking.
• To develop and evaluate a novel saliency model for predicting dashboard engagement.
• To derive design guidelines for effective dashboard visualization.

Main Methods:

• Conducted an eye-tracking study with 60 participants viewing 1,216 dashboards.
• Collected and analyzed eye-movement data to identify viewing patterns.
• Developed a dashboard-specific saliency model based on empirical data.

Main Results:

• Identified known and novel viewing patterns influenced by dashboard objects and layout.
• The developed saliency model outperformed state-of-the-art models for single visualizations.
• Demonstrated improved prediction accuracy for user viewing behavior on dashboards.

Conclusions:

• Dashboard design significantly influences user viewing behavior and engagement.
• The new saliency model offers a data-driven approach to dashboard design optimization.
• Findings can inform the creation of more effective and user-centered dashboards.