4a-2: Viz analysis

4 levels of validation

4 levels of validation: A framework

Domain - Data - Idiom - Tech

Visualization Analysis and Design, chapter 4, Munzner

Threats to validity

  • Wrong problem: You've misunderstood the needs
  • Wrong abstraction: You're showing the wrong thing.
  • Wrong idiom: The way you show doesn't work
  • Wrong algorithm: The code is too slow

Domain

Knowledge of an area that lets you frame problems from a human-centric interaction (HCI) design perspective.

This is similar to UX design thinking, especially the needfinding / problem definition portions.

In short, questions to validate why we are tackling the problems from the domain itself (Emphatize and Define).

Standford school of design thinking

UX design thinking tools

Tools to help with user empathy / problem specification / prioritization

  • Needfinding tools: Observations, interviews, surveys
  • Personas: User audience
  • Task analysis: Flowcharts and lists
  • Mindmaps: Clustering needs and prioritization

Data / task abstraction

What is the data like? Why are we doing the visualization?

Visual encoding / interaction

How are we designing the visualization?

Each of the main dataset types - tabular, networks, spatial (plus hierarchical) lead themselves to certain common visualization types.

Concepts like marks and channels, color theory, view manipulation (interaction), faceting will all be discussed in their own lectures.

Algorithmn / Technology

Execution of the visualization.

What technology stack are you using? Web? Desktop?

Are you using any 3rd party platform? Frontend stack? Backend stack?

Top-down or bottom up?

Typically either:

  • Designing a viz to solve a problem (top down)
  • Refining an existing solution (bottom-up)

Validation approaches

This is similar to the evaluation / testing phase of UX design thinking, where you iterate on a project.

More UX design thinking tools

Tools to help with user evaluation / testing

Case studies

Stories and experiences

Domain case studies

Maritime Maps / Transport Rail

Data case studies

Cybersecurity

Idiom case studies

Area to Boxplot

Tech case studies

Fashion start-up

Questions?

Chi-Loong | V/R