Paul Teetor - R Cookbook

Copyright 2011 Paul Teetor

OReilly books may be purchased for educational, business, or sales promotional use. Online editions are also available for most titles (.

Nutshell Handbook, the Nutshell Handbook logo, and the OReilly logo are registered trademarks of OReilly Media, Inc. R Cookbook , the image of a harpy eagle, and related trade dress are trademarks of OReilly Media, Inc.

Many of the designations used by manufacturers and sellers to distinguish their products are claimed as trademarks. Where those designations appear in this book, and OReilly Media, Inc. was aware of a trademark claim, the designations have been printed in caps or initial caps.

While every precaution has been taken in the preparation of this book, the publisher and authors assume no responsibility for errors or omissions, or for damages resulting from the use of the information contained herein.

Most recipes use one or two R functions to solve a specific problem. Its important to remember that I do not describe the functions in detail; rather, I describe just enough to solve the immediate problem. Nearly every such function has additional capabilities beyond those described here, and some have amazing capabilities. I strongly urge you to read the functions help page. You will likely learn something valuable.

Each recipe presents one way to solve a particular problem. Of course, there are likely several reasonable solutions to each problem. When I knew of multiple solutions, I generally selected the simplest one. For any given task, you can probably discover several alternative solutions yourself. This is a cookbook, not a bible.

In particular, R has literally thousands of downloadable add-on packages, many of which implement alternative algorithms and statistical methods. This book concentrates on the core functionality available through the basic distribution, so your best source of alternative solutions may be searching for an add-on package ().

The goal of every recipe is to solve a problem and solve it quickly. Rather than laboring in tedious prose, I occasionally streamline the description with terminology that is correct but not precise. A good example is the term generic function. I refer to print(x) and plot(x) as generic functions because they work for many kinds of x, handling each kind appropriately. A computer scientist would wince at my terminology because, strictly speaking, these are not simply functions; they are polymorphic methods with dynamic dispatching. But if I carefully unpacked every such technical detail, the essential solution would be buried in the technicalities. So I just call them functions, which I think is more readable.

Another example, taken from statistics, is the complexity surrounding the semantics of statistical hypothesis testing. Using the strict language of probability theory would obscure the practical application of some tests, so I use more colloquial language when describing each statistical test. See the for more about how hypothesis tests are presented in the recipes.

My goal is to make the power of R available to a wide audience by writing readably, not formally. I hope that experts in their respective fields will understand if my terminology is occasionally informal.