The thinner lines are important too, but vastly simplify to give the eye and easy comparison of the length of these lines across each plot, without the distracting and unnecessary horizontal lines. A center dot is given an extra thickness to point out the importance of the median value: the central value for each box/whisker plot. It’s about giving the viewer a better sense of the importance of each line. But we can do better, says Tufte! The bottom design isn’t just about saving ink. The top graphic shows the median and each quartile, which is useful information about the difference in averages and variance across different sets or parameters. One such example in the early pages of the book (I haven’t read the entire book yet) is a simple redesign of a box and whisker plot. How can we organize the data using shapes, line thickness, layout, size, and color to best deliver the important information to the viewer? Tufte is an expert in answering these questions and the 200 page books has examples from Leonardo da Vinci’s hand-drawn masterpieces of visual display of information in 1270 to the modern computer graphics renderings we appreciate today. Sometimes an analysis of a dataset may be expertly done, but poorly presented. Theory and practice in the design of data graphics, 250 illustrations of the best (and a few of the worst) statistical graphics, with detailed analysis of how to display data for precise, effective, quick analysis. The book explains the how-to of the display of data, rather than a how-to of the analysis. The Visual Display of Quantitative Information The classic book on statistical graphics, charts, tables. Tufte, who was has written several seminal books on the topic. It’s a classic, landmark book by the famous Edward R. I got the coolest Christmas gift: a book on the visualization of data. General Motors – Technical Education Program.Next Steps for Newly Admitted Doctoral Students.Next Steps for Newly Admitted Master’s Students.Box, line, and midgap plots: Effects of display characteristics on the accuracy and bias of estimates of whisker length. But they feel like they might not be as easy to read. (So if Tufte stops people doing that, then the more people who read him, the better).īut when there's tension between the two approaches then I'd choose the first (Cleveland).įor example, the Tufte (minimalist) boxplots manage to represent the same information as a box plot, but with less ink. They would certainly agree on the undesirability of most of the ridiculous stuff in the MS Excel plot menu. The conclusions of the two approaches are not necessarily incompatible. Or maybe like a philosopher trying to find the essence of a visualisation. Tufte feels a little like a designer trying to create beautiful visualisations based on a kind of minimalist aesthetic. But, very roughlyĬleveland feels more like a statistician trying to create visualisations which are efficiently and accurately perceived. The Cleveland books are not widely famous like The Visual Display of Quantitative Information, but I found them more appealing in a way that's kind of hard to describe. (And if you find that interesting, you would most likely enjoy the books mentioned above.) There's a classic paper by Cleveland and McGill, "Graphical Perception: Theory, Experimentation, and Application to the Development of Graphical Methods" (you can download a PDF) which is also interesting. The Elements of Graphing Data (1ed 1985, 2ed 1994)Īfter seeing references to Cleveland in the R documentation (for example, the loess and lattice packages), I read both the Cleveland books, and found them extremely interesting. My favourite would be (both) Bill Cleveland's books R's powerful ggplot2 graphics system has a default output style which follows many of these principles, and it looks good.īut it's not my favourite book in this area.
0 Comments
Leave a Reply. |