I've just finished reading The Visual Display of Quantitative Information, by Tufte.
Roughly half of the book is dedicated to an interesting analysis of the origins and history of charts. This includes extensive surveys of charts from widely different (in epoqué, scope, topic) publications, to display examples of both good and bad charts.
The second half is devoted to a theory —a small set of guiding principles— on how to produce useful graphs to explain data. The text continues to show interesting examples and revisits a few common types of graphs to suggest some generic improvements that can be applied to a large number of graphs.
Before I started reading the book, I was expecting it would contain a complete and in-depth catalogue of interesting techniques for displaying different types of data, but the theory falls way short of that. The principles are as basic as “Above all else, show the data”, “Erase non-data-ink” or “Revise and edit”. I don't have anything against simple principles, but these are far more generic and obvious, less concrete, than what I was expecting.
All in all, it's a relatively “light” book, that you'll probably read in just a few hours. It is well written and pleasant to read. You'll probably get a few interesting ideas out of it. Not a bad book, but not specially good either.
Discussion (2)Last update: 2010-01-10 (Rev 16612)