Visualization Design

What is the purpose of visualization design?

To move systematically from a visualization problem to a visualization solution by understanding data, tasks, constraints, and iterating through structured design processes.

What is the DIKW hierarchy and why is it relevant?

It describes the progression from Data → Information → Knowledge → Wisdom. Visualizations often aim not to show raw data but to help extract information from it.

What is the difference between visualizing data vs. visualizing information?

Raw data visualizations (e.g., plain parallel coordinates) often overplot and hide structure, whereas information visualizations apply transformations (e.g., bundling) to reveal patterns.

What is the purpose of data abstraction?

To describe data independently of domain, enabling expressive visualizations that show the data and only the data.

What is the difference between data context and data content?

Data context = reference space/independent variables (e.g., location, time)

Data content = measured values/dependent variables (e.g., temperature, wind)

What is the “extent of validity” in the data context?

How far a measured value remains valid (point, narrow local, broad local, global) and whether interpolation is meaningful.

What is Shepard’s method (inverse distance weighting)?

An interpolation method that estimates unknown values by weighting nearby observations inversely by distance, controlled by parameter p.

How does the power parameter p influence interpolation in inverse distance weighting?

Low p: smoother interpolation

High p: sharp boundaries approaching Voronoi-like regions

What topology types can data be organized in?

Grid-based (regular, irregular) and graph-based (network, hierarchy). The topology affects valid operations and interpolation.

What are scalar, vector, and tensor data?

Scalar = one magnitude

Vector = magnitude + direction

Tensor = multiple magnitudes/directions; requires specialized visualization

Why do we perform task abstraction?

To describe analytic tasks independently of domain, supporting effective (task-aware) visualization choices and requirements engineering.

What is a task in the visualization context?

A mid-level abstraction between goal (why) and interaction actions (how) - e.g., “overview”, “zoom/filter”, “details-on-demand”.

What is the difference between unipolar and bipolar tasks?

Unipolar tasks have no natural opposite (e.g., validate); bipolar tasks come in oppositional pairs (add/remove, locate/identify, undo/redo).

What is the difference between identify and locate tasks?

Identify = given a position, find the value

Locate = given a value, find where it occurs

What are synoptic tasks?

Tasks that describe sets of items (e.g., trends, distributions), rather than individual data points - basically, they’re extensions of elementary tasks.

Give an example of synoptic lookup vs. elementary lookup.

Elementary: “What was the value on this date?”

Synoptic: “What was the trend during this interval?”

What is a relation-seeking task?

A task where the user seeks pairs/sets of items satisfying a specified relation (e.g., “When did the price triple?”)

What do we gain by combining data and task abstraction?

A full statement of the visualization problem → domain-independent requirements → ability to search for fitting techniques.

What is the Design Activity Framework?

A four-stage process: Understand → Ideate → Make → Deploy, guiding the entire visualization design workflow.

What is the purpose of the “Understand” stage in the Design Activity Framework?

To derive visualization requirements by studying users, data, tasks, constraints.

Outcome: a clear visualization problem specification.

What is the purpose of the “Ideate” stage in the Design Activity Framework?

To generate many externalized visualization ideas (sketches, mockups) that satisfy the requirements.

What is the purpose of the “Make” stage in the Design Activity Framework?

To turn selected ideas into hi-fi prototypes suitable for user feedback before committing to full implementation.

What is the purpose of the “Deploy” stage in the Design Activity Framework?

To create a fully functional visualization system that works with real data in real settings using software engineering practices.

What is the 5 Design Sheet Method used for?

For ideation - systematically developing, combining, and selecting visual ideas across 5 structured sheets.

What happens on Sheet 1 of the 5DS method?

Brainstorm 15-20 ideas, filter them, cluster similar ideas, combine/refine across clusters, then choose 3 to continue with.

What do Sheets 2-4 contain in the 5DS method?

Three detailed alternative designs, each with: information section, layout sketch, interactions, focus point, pros/cons.

What is Sheet 5 in the 5DS method?

The final design realization with algorithms, dependencies, cost estimates, hardware requirements.

What are the four guiding design principles?

Functional design, deliberate design, intuitive design, ethical design - ensuring usefulness, purpose, simplicity, and honesty.

What is Munzner’s Nested Model?

A model describing four design levels (domain problem, data/task abstraction, encoding idiom, algorithm) and how to evaluate issues at each level.

What evaluation methods correspond to the Nested Model levels?

Domain: interviews, field studies

Abstraction: check task/data alignment

Encoding/Interaction: user studies

Algorithm: performance benchmarks, image-based metrics (e.g., overplotted %, edge crossings).