You are here

Visualization Analysis and Design

Tamar Munzner
Publisher: 
CRC Press
Publication Date: 
2015
Number of Pages: 
404
Format: 
Hardcover
Series: 
A K Peters Visualization Series
Price: 
79.95
ISBN: 
9781466508910
Category: 
Monograph
[Reviewed by
William J. Satzer
, on
03/17/2015
]

This is a book about visualization, and specifically the presentation of data in visual form. One can think of visualization as a kind of transformation that takes information in a form suitable for computation into forms more compatible with human perception and cognition and more easily communicated. The current book is an introduction to the subject. It is aimed at graduate students and advanced undergraduates in computer science, but it would be accessible to a much broader audience.

Visualization is a young field with only a few primary texts available. What distinguishes this one is the way it synthesizes past work to develop a comprehensive framework for design and analysis. Edward Tufte’s books, especially Envisioning Information, were eye-opening to many of us involved in preparing data for analysis, talks, reports and papers. But Tufte concentrated on static design presentations and there are now many new possibilities for interactive and dynamic graphical data analysis and presentation.

The author builds a framework for understanding the key elements of visualization and develops a synthesis of current best practices. She breaks down the framework for analysis into three steps: what, why and how. “What” addresses the kind of data that is to be visualized; “why” identifies the purposes for creating the visualization; “how” describes the method by which it is carried out. In this book, the “what” includes three generic data types: tables, spatial data, and networks. The table category includes scatterplots, bar charts, dot and line charts, heat maps and pie charts. Maps, vector, tensor and flow fields are spatial data. Networks include tree structures and adjacency matrix views.

The first part of the book addresses the what, why and how questions in detail and then considers the validation question: how well does a particular visualization actually work? This leads naturally into a discussion of visual channels; these are the means of controlling the basic graphical elements in an image. Channels include spatial position, length, angle, area, depth, color, curvature, volume, shape, and motion. Matching the channel to the human perceptual system is a key to effective visualization design. As a simple example, think of pie charts and bar charts. Pie charts require both angle and area judgments, which are perceptually harder and less accurate than bar charts that require only judgments of length on a common scale.

One of the most useful chapters to a casual reader is on rules of thumb for visualization. These include: don’t use 3D when 2D will do, and don’t use 2D when a one-dimensional list will do. Using our eyes to switch between two views that are visible simultaneously has a much lower cognitive load than relying on memory to compare a current view with one seen before… There are about half a dozen similar rules presented in the book. They are perceptive and clearly based on years of experience.

The book is filled with examples from the universe of visualization. Virtually all the possibilities for visualization design are illustrated with specific examples.

It’s also worth pointing out what’s not in the book. The author intentionally takes a top-down look at visualization. This does not extend all the way to the level of algorithms. As the author notes, the book is already pretty long and would need to double in size to incorporate an adequate treatment of algorithms. On a more fundamental level, the book is about visualization of data, not about visualization in the broader sense that mathematicians sometimes use. So there is no visualization of the hypercube here, no eversion of the sphere, not anything like that.

This is an attractive book, one that’s likely to be a fundamental source for the field. It’s worth a look for anyone with even a passing interest.


Bill Satzer (wjsatzer@mmm.com) is a senior intellectual property scientist at 3M Company, having previously been a lab manager at 3M for composites and electromagnetic materials. His training is in dynamical systems and particularly celestial mechanics; his current interests are broadly in applied mathematics and the teaching of mathematics.

What's Vis, and Why Do It?
The Big Picture
Why Have A Human in the Loop?
Why Have A Computer in the Loop?
Why Use An External Representation?
Why Depend on Vision?
Why Show The Data In Detail?
Why Use Interactivity?
Why Is the Vis Idiom Design Space Huge?
Why Focus on Tasks?
Why Focus on Effectiveness?
Why Are Most Designs Ineffective?
Why Is Validation Difficult?
Why Are There Resource Limitations?
Why Analyze?

 

What: Data Abstraction
The Big Picture
Why Do Data Semantics and Types Matter?
Data Types
Dataset Types
Attribute Types
Semantics

 

Why: Task Abstraction
The Big Picture
Why Analyze Tasks Abstractly?
Who: Designer or User
Actions
Targets
How: A Preview
Analyzing and Deriving: Examples

 

Analysis: Four Levels for Validation
The Big Picture
Why Validate?
Four Levels of Design
Angles of Attack
Threats and Validation Approaches
Validation Examples

 

Marks and Channels
The Big Picture
Why Marks and Channels?
Defining Marks and Channels
Using Marks and Channels
Channel Effectiveness
Relative vs. Absolute Judgments

 

Rules of Thumb
The Big Picture
Why and When to Follow Rules of Thumb?
No Unjustified 3D
No Unjustified 2D
Eyes Beat Memory
Resolution over Immersion
Overview First, Zoom and Filter, Details on Demand
Responsiveness Is Required
Get It Right in Black and White
Function First, Form Next

 

Arrange Tables
The Big Picture
Why Arrange?
Classifying Arrangements by Keys and Values
Express: Quantitative Values
Separate, Order, and Align: Categorical Regions
Spatial Axis Orientation
Spatial Layout Density

 

Arrange Spatial Data
The Big Picture
Why Use Given?
Geometry
Scalar Fields: 1 Value
Vector Fields: Multiple Values
Tensor Fields: Many Values

 

Arrange Networks and Trees
The Big Picture
Connection: Link Marks
Matrix Views
Costs and Benefits: Connection vs. Matrix
Containment: Hierarchy

 

Map Color and Other Channels
The Big Picture
Color Theory
Colormaps
Other Channels

 

Manipulate View
The Big Picture
Why Change?
Change View over Time
Select Elements
Navigate: Changing Viewpoint
Navigate: Reducing Attributes

 

Facet into Multiple Views
The Big Picture
Why Facet?
Juxtapose and Coordinate Views
Partition into Views
Superimpose Layers

 

Reduce Items and Attributes
The Big Picture
Why Reduce?
Filter
Aggregate

 

Embed: Focus+Context
The Big Picture
Why Embed?
Elide
Superimpose
Distort
Costs and Benefits: Distortion

 

Analysis Case Studies
Graph-Theoretic Scagnostics
VisDB
Hierarchical Clustering Explorer
PivotGraph
InterRing
Constellation

 

Bibliography

 

Further Reading appears at the end of each chapter.

Dummy View - NOT TO BE DELETED