Stephen Few introduces the visual analysis of data. He shows readers how to discover patterns in large data sets through clever arrangement, highlighting and filtering of data points. I encountered the book as the text in a four-week online class on visual data analysis. But it also works well as a standalone introduction to this area.
The first half of the book has a different focus than I expected. Few suggests that "...we've largely ignored the primary tool that makes information meaningful and useful: the human brain. While concentrating on the technologies, we've forgotten the human skills that are required to make sense of the data." He describes the human visual system, how it processes information, and the errors in perception it sometimes makes. His emphasis, however, is on the strengths of visual perception which he links to best practices in data analysis. One of the most useful parts of this section is in Chapter 2, where he lists and describes the "aptitudes and attitudes of effective analysts."
The book's second half describes and illustrates specific visual analysis techniques. It is rich with visual examples, comparisons of effective and ineffective displays, and series of related visualizations which show incremental steps of data transformation and analysis. Chapters are organized by specific data patterns and analytical techniques, describing how to look for the following six kinds of patterns:
- Time-series
- Ranking and part-to-whole relationships
- Deviations
- Distributions
- Correlations
- Patterns in multivariate data
Two final chapters present recommendations for developers of data analysis software and make predictions about future trends in visual data analysis.
The book is recommended for any researcher who works with large data sets. It is well-written, contains clear examples, and references recent research and the latest tools available for data analysis. Readers may also be interested in Few's
Show Me the Numbers: Designing Tables and Graphs to Enlighten which discusses how to best describe patterns in data to nonresearchers.