Intro (Ch 1), Visualization Analysis & Design, 2021

Introduction

In this section, Tamara Munzner introduces herself and the topic of visualization analysis and design. She provides a definition of visualization and explains why it is important to have humans in the loop.

• Visualization systems provide visual representations of data sets designed to help people carry out tasks more effectively.

• Visualization is suitable when there's a need to augment the capabilities of a human, rather than simply replacing people with purely computational decision making methods.

• There are many situations where fully automatic solutions do not exist or are not trusted, such as ill-specified analysis problems or situations with many branching paths.

• Visualization replaces cognitive acts with perceptual acts, freeing up cognition for higher level questions.

Situations Where Visualization is Useful

In this section, Tamara Munzner discusses different situations where visualization can be useful.

Long-term Goal

• Exploratory visual data analysis is a common use case where people will continue doing it indefinitely.

Presenting Known Results

• Data journalism often involves presenting known results.

• People may present reports about what they've discovered in finance or teaching.

Developing an Automatic Solution

• Visualization might be used as a stepping stone towards something that is more automatic or supported by some automation.

• Visualization might be aimed at algorithm or model developers to help them refine some sort of computational solution.

• Even if there's already a fully automatic solution, visualization might still be needed to build trust or monitor the solution.

Why Use Visual Representation?

In this section, Tamara Munzner explains why visual representation is important.

• With visualization, we're trying to replace cognitive acts with perceptual acts in order to free up cognition for more interesting, typically higher level questions.

• Visual encoding replaces the cognition with perception, making it easier to understand overall trends and patterns in data.

The Importance of Vision in Visualization

In this section, the speaker explains why vision is crucial for visualization and how the human visual system is an extraordinarily high bandwidth channel into our heads.

Why Vision?

• The human visual system is an extraordinarily high bandwidth channel into our heads.

• We have evolved to do an enormous amount of essentially background processing, which makes us think we see everything around us all simultaneously.

• This capability of getting an overview of things is completely crucial for a lot of how we use visualization.

Other Sensory Modalities

• Touch and haptics - right now, we don't have very good ability to record or replay.

• Taste and smell are not yet really viable for recording and playing.

Representing Data with Visualization

In this section, the speaker explains why visualization is important in representing data and how it can help you see more rich detail than summarized information.

Importance of Visualization

• Summaries are crucial but they do lose information. Visualization can help you both confirm what you expect to see and find something unexpected.

• Anscombe's Quartet is a great example where identical simple descriptive statistical descriptors result in different plots that show dramatic differences.

• Visualization helps check if the data underlying a statistical model matches appropriately.

Complex Heterogeneous Data Sets

• When dealing with complex heterogeneous data sets, it's important to try to see them through visualization.

Resource Limitations in Visualization Design

In this section, the speaker discusses three different kinds of resource limitations that visualization designers must take into account: computers, humans, and displays.

Types of Resource Limitations

• Computational limits include CPU time and system memory.

• Display limits are often the most constrained resource in visualization due to limited pixels available for display.

• Human limitations include how long it takes a person to complete a task, how much information they can remember, and how long they can focus on something.

Information Density

• Information density is the ratio between the space used to visually encode information versus all of the white space that is unused.

• Finding the sweet spot between density and sparsity allows for effective communication while still seeing things in enough detail.

Benefits of Visualization Analysis and Design

In this section, the speaker explains why analyzing existing visualizations is useful for designing new ones.

Imposing Structure on Design Space Possibilities

• Analyzing existing visualizations provides a scaffolding to hang thoughts on when designing new ones.

• Most possibilities of visually encoding and interacting with data are not effective, so analyzing existing visualizations helps guide search.

Example: Visualizing Large Trees

• Two systems for visualizing large trees are compared: SpaceTree and TreeJuxtaposer.

• Both systems use visual encoding and navigation through selection to find paths from leaf nodes up to ancestors.

• SpaceTree filters out certain parts of the tree based on user selection while TreeJuxtaposer rearranges the tree with stretch and squish navigation.