2cultures.net(.au)

Alberto Cairo's newly translated book on information graphics, The Functional Art, is a healthy mix of theory and how it applies in practice, and much of it comes from Cairo's own experiences designing graphics for major news publications. (I don't think Alberto remembers, but what seems like many years ago, I sat right behind him for two weeks at the New York Times when they brought him in to help illustrate Raphael Nadal's approach to tennis.)

His experience is hugely important in making the book work. There's a growing number of books on information graphics, and many are written and illustrated by people who don't have much experience displaying information, which leads to art books posing as something else. This isn't one of those books. Cairo knows what he's talking about.

As you flip through, you'll notice a lot of examples, with a focus on process and even a handful of pencil sketches. The last third of the book is interviews with those well-established in the field, which also walks you through how some graphics were made. There's a strong undertone of finding the balance between function (e.g. efficiency and accuracy) and engagement (e.g. use of circles).

Cairo comes from a journalism background, so the book is mostly in the context of presentation, but there's of course plenty that you can apply to more exploratory graphics. I would say though that Cairo's strength is in illustration and information, and so the book reflects that. This isn't a book that covers visual data analysis or statistical concepts, but it is one that explores and describes the making of high quality information graphics that lend clarity to concepts and ideas. If you're looking for the latter, The Functional Art is worth your time.

Check out the sample chapter on the publisher page, but then grab it on Amazon and save a few bucks.

R, the favorite computing language of a growing number of statisticians, is friendly enough that you can get a lot done without being an expert programmer, because there are a lot of packages and built-in functions that can take care a lot of the grunt work for you. Learn how to use a function, prepare your data, and you get some output. However, as you use R more, whether it's for analysis or just for graphics, there comes a point when there isn't a package or function that does exactly what you want.

Norman Matloff's Art of R Programming is for those who want to learn to write their own software in R. This is an R programming book that starts from the beginning — running R, vectors, lists — to the more advanced such as simulations, object-oriented programming, and debugging.

As the back cover says, no statistical knowledge is required to learn from this book, because it really is a book about programming written by a computer science professor (although Matloff is a former professor of statistics).

The technical reviewer and an active member of the R community, Hadley Wickham no doubt provided additional depth to the text. There are lots of examples and explanations to go along with the code.

The graphics chapter is relatively basic, explaining how to produce a few plots, so you won't get a ton out of this part other than a foundation to look into other resources. If you really want to learn the nuts and bolts of graphics in R, Paul Murrell's R Graphics is a good choice (but kind of pricey). And of course Visualize This for more high-level material. I only mention this though because this is a visualization blog — not because I was expecting more about graphics in Art of R Programming.

Here's who Matloff says the book is for:

Many use R mainly in an ad hoc way—to plot a histogram here, perform a regression analysis there, and carry out other discrete tasks involving statistical operations. But this book is for those who wish to develop software in R. The programming skills of our intended readers may range anywhere from those of a professional software developer to "I took a programming course in college," but their key goal is to write R code for specific purposes.

That's about right. If you've never written a line of code, you might find some of the concepts challenging, but if you have at least a vague idea of what programming is, you should find Art of R useful. I'm keeping this one.

An open-source, "rough and partial" draft of the book can be found here [pdf]. The draft has a lot of the same ideas so you can get a good idea for what Art of R Programming is about, although from a quick scan, the final is more polished and contains more material.

R can be confusing when you're first starting out, especially when you don't have any experience in programming. There's a lot of documentation online, and package developers do a decent job at providing examples on how to use their work in your code, but that stuff is not always easy to find. It's easy if you know the name of the package or function you're looking for. However, most of the time you just know what you want to do—like sort a data frame or test a regression model—and not the name of a package.

The R Cookbook by developer Paul Teetor might be your answer.

Overview

From the book description:

This collection of concise, task-oriented recipes makes you productive with R immediately, with solutions ranging from basic tasks to input and output, general statistics, graphics, and linear regression.

There are 14 chapters and around 400 pages. One third covers the basics of R, such as setting variables; one half covers analysis tools such hypothesis testing and linear regression; the rest covers miscellaneous topics. There's one chapter on creating graphics.

The Insides

Those who have used other O'Reilly cookbooks will recognize the format right away. Sections are organized by recipe with a problem, solution, and discussion. Most recipes are pretty short, around one or two pages.

In short, R Cookbook is basically what I expected. This is a good thing, as O'Reilly cookbooks are usually pretty useful. The recipes are straightforward to follow and the text is an easy read. There's some discussion at the beginning of each chapter about statistical methods, such as what p-values mean, but don't expect a full-on guide on statistical analysis (not that it claims to be one).

What this book will provide are steps that can help you with the early stages of getting up and running to the more advanced functions for probability, general statistics, and time series analysis.

Bottom Line

For those who already use R, the R Cookbook can be a handy reference when Google lets you down. I imagine those who are familiar with statistical methods but use different software like SAS will also find this useful. If, however, you're looking for a book that's more visualization-based or you're new to statistics, you will probably want to look elsewhere.

[Amazon Link]