Customer Reviews

ggplot2: Elegant Graphics for Data Analysis (Use R!)

5 star:		(12)
4 star:		(6)
3 star:		(2)
2 star:		(1)
1 star:		(0)

 

 

ggplot2 is an R package for creating graphs. If you have used R, you may have observed that you can control almost all aspects of a graph you create. This flexibility can come at a price in having to be familiar with several options and syntax that is not always intuitive. ggplot2 overcomes this hurdle because it is based on a grammar. After one invests the time to learn this grammar, one can create customizable plots with default settings that are aesthetically pleasing. ggplot2 automatically takes care of details such as legends. This book covers ggplot2 plotting in a clear manner. It is filled with examples, simple and complex, and all code is available from the author's website. It covers the logic of the grammar, basic plots, combining plots, how to represent subsets, and even how to format data so that it can be plotted. I highly recommend this book. You will produce plots exactly as you want them (with ease), and learn about different ways to display your data. Be sure to invest some time up front to develop an understanding rather than cutting and pasting code. This investment will pay off.

This book is a practical overview of ggplot2, a new graphics package available for R - and I would argue the best statistical graphics package for any language. Why? Because it allows users to create gorgeous, elegant graphics with a minimum of effort.

As Wickham details in this lucid book, ggplot2 is based on a grammar of graphics, which conceives of statistical plots not as fixed chart typologies (bar, line, pie) to be filled with data, but as graphical compositions built up from component parts (scales, coordinate systems, geometric objects).

Before embarking on several chapters describing how to control various elements of the grammar, Wickham dives in (with Chapter 2, "Getting started with qplot") and guides users on how to rapidly create graphics with the qplot (for "quick plot").

Two elements struck me in reading this book and beginning to use ggplot2:

(i) It possesses excellent defaults. Wickham has put an enormous amount of thought into the default aesthetics. From light gray plotting backgrounds to monochromatically spaced color palettes, ggplot2 builds on decades of work done in the information visualization space.

(ii) It possesses an intuitive syntax. Creating graphics in R often requires specifying an enormous number of parameters. ggplot2's syntax - the ability to "build up" a graph from layer by layer - allows for an intuitive way to control complexity.

If you are serious about statistical graphics -- and even if you are new to R -- this book is an essential volume.

Inspired by Wilkinson (2005), Wickham has developed and implemented ggplot2, a layered grammar of graphics that makes intuitive the production of meaningful (and frankly beautiful) data representations. This book walks the reader through two modes of using ggplot2. For the novice seeking quick default depiction of data, the qplot() function is introduced as an effective panacea of simple plotting problems. For the user seeking to achieve more advanced graphics than possible with qplot(), Wickham introduces his layered grammar and a host of functions that simplify application of this grammar to common graphing problems. No matter whether one employs the qplot() or more advanced graphic generation approaches, the result is a printable, save-able, and highly customizable graphic object. Wickham provides extensive customization examples, including the development of themes that can be applied across multiple individual graphics, simplifying the creation of custom yet aesthetically unified graphics. As a student of cognitive science, this reviewer is impressed by the work Wickham has done to ensure that ggplot2 and, through this book, its users are sensitive to important psychological aspects of generating graphics that link abstract data and statistical interpretation via the human visual perceptual system. The result of this attention to detail is a graphing package that not only lets you plot data easily and meaningfully, but also beautifully. This book is an exemplary resource for learning the power of ggplot2 and as such is recommended to all.

Hadley Wickham does for graphics what formula notation [1] does for regression in R: Shorten the way from concept to code. With ggplot2, graphics code becomes easier to read, maintain and reuse. The "grammar of graphics" identifies the role that each variable plays in the display (e.g. response, predictor), how to summarize it (raw data, histogram, smoothing, ...), and what visuals to present it with (called "geoms": line, point, area...). This takes some getting used to, but qplot() makes a familiar starting point [2].

As evident from the ggplot2 website and mailing list, this package is a solid piece of work. The book is an excellent overview and introduction to ggplot2. In my opinion it is very well balanced, with enough examples to stimulate creativity, enough detail to explain principles, and concise enough to be very readable. I had downloaded earlier drafts of the book, but still found it worthwhile to buy a hardcopy.

[...]

ggplot2 is a remarkable accomplishment by its author, but I am not as happy with the companion book. I shudder to think of mastering ggplot2 without it (a definite "buy", then), but, disappointingly, even with it things are not as clear as they ought to be.

The general idea of the grammar of graphics is very intriguing, but the book has a lot of code to produce diagrams with relevant information coded in colours - unfortunately all diagrams are printed in b&w! My favourite is a colour wheel printed completely in grey ;)
I payed more than 50 Euros for this 200-page book and I think for this price Springer could add some colours!

The author thoroughly describes the basics of his R implementation of the grammar of graphics and it would have been hard to find this information elsewhere. But even if the cover page says "you will need some basic knowledge of R (i.e. be able to get your data into R)" it is definitely useful to know somewhat more to be able to follow the examples.

There are some small flaws in the book which I would expect to be gone in the 2nd edition, i.e. the missing geom_rect in the table 4.2 of ggplot geoms.

Because this is a book about a specific package in R, I will try to avoid ascribing qualities of the package to the book and vice versa. In order to help with that I should mention at the gate that ggplot2 is a fabulous package. The structure is sensible, the options are manifold and support is great.

The book itself is helpful, but not exhaustive. I think this was a deliberate choice on the part of the author--he didn't want to load down potential users and developers with a complete codex of options, parameters and strategies. This is in keeping with the philosophy of the addon. Give users a solid framework and (largely) "get out of [their] light". Full color (or colour!) examples abound, another pleasant change from books in this series. If a technique involves color then the figure will be displayed immediately and not relegated to a series of color plates. We are given a quick tour through the notion of a grammar of graphics and then set loose to play with customizing various plots. Attention is paid to showing plot types which are appropriate for the underlying data structures.

However the book is not meant to serve as a desk reference. As you will find when you explore the ggplot2 package, many options are "hidden", insofar as they are either shared by scales or aesthetics or simply not recoverable through the r help (e.g. type ?scale_y_continuous and you'll see what I mean). So mid level users who have moved beyond qplot but aren't yet experts in the vagaries of grid or grobs are left in the reeds. A few dozen times I have wished for an omnibus reference where I could look for all possible options for rendering legends, combining data frames on one plot or dealing with axes. This book will not provide such a reference (and the online documentation on Hadley's website, while great, is a bit incomplete).

However, that is a relatively minor concern. If you are new to ggplot2 and want a great course on learning more about the package and the grammar of graphics, this book is for you.

ggplot2 is simply fantastic. R already has great base graphics, and the lattice package allows one to do some great things. However, the ggplot2 system provides an intuitive and simple way to create complex plots. The system encourages users to think about the elements of a plot rather than conceiving of a plot as just a bunch of settings. Ggplot2 plots, even using default settings, are usually near-publication quality and only require small tweaks (if any) to get to a polished, publishable result. But this is a review of the book, not the R package (almost any seasoned R user will tell you that ggplot2 is worth learning).

The author has provided great online resources for learning how to use ggplot2, and one could gain a fairly good understanding of the package by using the online tutorials. However, the book provides a perfect desktop reference--something to thumb through when you aren't sure exactly what you need to do next. GGplot2 is based on an innovative paradigm called "the grammar of graphics." **Reading the book will help users understand not only the technical aspects of ggplot2, but also the paradigm on which it is based: I, for example, find myself thinking much more clearly about what plots are, and which plots are best to accomplish a given goal.** Further, sometimes a hard-copy, printed book just can't be beat, especially if one wants to use the book as a tutorial, working from front to back, as is my preference.

It is worth noting that Hadley Wickham (the author) is one of the R community's greatest assets. In addition to ggplot2, he has created numerous other R packages (such as plyr and reshape, both of which are indispensable) and he spends quite a bit of time online supporting users (e.g. in stackoverflow). R simply would not be the same without him, and I don't hesitate for a moment to support his work by purchasing this book (the man has saved me countless hours with his plyr package alone!).

This is a terrific guide to ggplot2. A clear introduction to what can be a very complicated topic. However the kindle edition of this book is terrible, no links in the index for example so if you need to look something up you have to guess as to the page number. Buy the physical book.

Excellent book for my style of learning! The book is broken out logically into short chapters. Each chapter contains simple, self-contained code examples, with pictures of the graphics generated by that code. It is well-indexed, but the graphics themselves serve as the best index to the information. I worked through the book, chapter by chapter, typing in the code examples, and was able to generate the graphics by myself, which made the information very easy and accessible. The examples are from publicly available data sets, so I didn't have to create my own data to test out the information. I also picked up a lot of useful information about programming in R from the style of the coding, which also improved my knowledge about R overall. I am a good programmer, although I was a beginner in R when I started this book. This book significantly improved my knowledge of this very powerful, but somewhat arcane, language.