Same Data, Different Audiences: Using Personas to Scope a Supercomputing Job Queue Visualization

Area of Science:

• Data Visualization
• Human-Computer Interaction
• Scientific Computing

Background:

• Domain-specific visualizations often limit utility to a single user group.
• Designing for multiple user groups presents challenges and opportunities for broader applicability.
• Balancing scope and specificity is key to enhancing visualization utility.

Purpose of the Study:

• To develop and evaluate Guidepost, a versatile notebook-embedded visualization.
• To support diverse tasks across scientists, machine learning researchers, and system maintainers.
• To explore the use of personas for categorizing tasks in visualization design.

Main Methods:

• Developed Guidepost, a visualization integrated within a notebook environment.
• Adapted persona-based design to categorize tasks by stakeholder uniqueness.
• Utilized interactive visualizations for shared tasks and scripting for unique tasks.
• Evaluated Guidepost through real-world case studies and user evaluations.

Main Results:

• Guidepost effectively supports distinct tasks for scientists (compute wait times), machine learning researchers (prediction accuracy), and system maintainers (usage trends).
• The visualization's visual encodings, interactions, and export capabilities cater to varied user needs.
• Persona-based task categorization guided the design for multi-group support.

Conclusions:

• Guidepost demonstrates the value of a unified visualization approach for diverse data analysis needs.
• Integrating interactive visualizations with scripting offers a flexible solution for complex data tasks.
• Persona-driven design is effective for creating visualizations that serve multiple stakeholder groups.