Monday, February 25, 2013
Show Me the Numbers
I have been fully reading Stephen Few's book: Show me the Numbers: Designing Tables and Graphs to Enlighten. (Second Edition). In part for preparation of some data visualization design choices, and also because the topic of linking strong, accurate and well designed data visualization to business decisions has always been an interest of mine.
First of all bravo to this book. Beautifully written and formatted. An excellent and easy to read book on the subject. I rarely recommend reading such a 'how to' book in its entirely. In this case it is well worth the actual time. Few's teaching capability comes across quickly.
Also notable is his immediate linkage of visualization to analytics. His chapter 2 outlines the simple statistics most often encountered in such work. It's particularly useful to those without formal stat backgrounds. It is by no means all the analytics the user should understand, but is an essential start. The clear exposition that will get you 90% of what you need to support visualization.
Also unusual is the book's inclusion of an examination of the comparative use of tables and graphs. Depending on the decisions being made, one or the other form of data display can be best to deliver your point clearly. Its all about using the simplest design possible to minimize the work needed to get results. It is remarkably similar to user interface design. Few does not show designs for specific computer packages, but makes the case that many of the simpler approaches can be produced by packages like MS Excel. He does criticize Excel as introducing too many obscuring concepts. He also shows a number of Tableau Software visualization examples.
Later in the book he discusses the role of visual perception in graphical communications. Then he looks more broadly at variations in graphical design, including geographical information examples. Also an excellent section on the neglected topic of component level graphical design. Lots of illustrations too, which makes the latter chapters easy to scan for examples. The book does not compare business intelligence packages broadly, but deals with the design of data visualization.
A mild criticism is that the book does not deal with the process of end-user interaction design with data visualizations. Perhaps that is another volume. I would like to see that examined further.
Later chapters include thoughts about how to tell compelling stories with numbers. And also the interaction of standards and innovation in the visualization space.
Read it, it is very good. See also Stephen Few's blog which covers related topics.