The purpose of these notes is to provide a quick introduction to R, particularly as a tool for fitting linear and generalized linear models. The notes originated in a 1992 handout introducing S-Plus, but have been substantially rewritten to target R using R Studio.
Each of the subjects discussed here includes a sample application. Additional examples will be found in the R logs of my GML course, as well as in the pages for courses on Multilevel Models, Survival Analysis and Demographic Methods, all at https://data.princeton.edu.
R is a powerful environment for statistical computing which runs on several platforms. These notes are written specially for users running the Windows version, but most of the material applies to the Mac and Linux versions as well.
R was first written as a research project by Ross Ihaka and Robert Gentleman, and is now under active development by a group of statisticians called ‘the R core team’, with a home page at https://www.r-project.org.
R was designed to be ‘not unlike’ the S language developed by John Chambers and others at Bell Labs. A commercial version of S with additional features was developed and marketed as S-Plus by Statistical Sciences, eventually becoming part of TIBCO Spotfire. You can view R and S-Plus as alternative implementations of the same underlying S language. The modern R implementation, however, is by far the most popular.
R is available free of charge and is distributed under the terms of the Free Software Foundation‘s GNU General Public License. You can download the program from the Comprehensive R Archive Network (CRAN). Ready-to-run ’binaries’ are available for Windows, Mac OS, and Linux. The source code is also available for download and can be compiled for other platforms.
These notes are organized in several sections, covering getting started, reading data, fitting linear models, and fitting generalized linear models. I close with a few pointers and references on where to go from here. I have tried to introduce key features of R as they are needed by students in my statistics classes. As a result, I often postpone (or altogether omit) discussion of some of the more powerful features of R as a programming language.
S was first introduced by Becker and Chambers (1984) in what’s known as the ‘brown’ book. The new S language was described by Becker, Chambers and Wilks (1988) in the ‘blue’ book. Chambers and Hastie (1992) edited a book discussing statistical modeling in S, called the ‘white’ book. The latest version of the S language is described by Chambers (1998) in the ‘green’ book, but R is largely an implementation of the versions documented in the blue and white books. Chamber’s latest books, Chambers (2008) and Chambers (2016), focus on programming with R.
Venables and Ripley (2002) have written an excellent book on Modern Applied Statistics with S-PLUS that is now in its fourth edition. The latest edition is particularly useful to R users because the main text explains differences between S-Plus and R where relevant. A companion volume called S Programming appeared in 2000 and applies to both S-Plus and R (Venables and Ripley 2000). These authors have also made available in their website an extensive collection of complements to their books, follow the links at MASS 4.
There is now an extensive and rapidly growing literature on R. Good introductions include the books by Krause and Olson (1997), Dalgaard (2008), and Braun and Murdoch (2016). Beginners will probably benefit from working through the examples in Hothorn and Everitt (2014) A Handbook of Statistical Analyses Using R, now in its third edition, or Fox (2002)’s companion to applied regression. Among more specialized books my favorites include Murrell (2006), an essential reference on R graphics, Pinheiro and Bates (2000), a book on mixed models, and Therneau and Grambsch (2000)’s book on survival models. (Therneau wrote the survival analysis code used in S-Plus and R.)
Hadley Wickman has made a number of ground-breaking contributions to R that deserve special mention.He is the author of the
ggplot2 package (Wickman 2016), a very popular graphics package that has brought to R the grammar of graphics proposed by Wilkinson (2005). He has also contributed a number of data-management packages, of which the most notable is
dplyr, and has advocated the use of “tidy” datasets. His approach to data management is explained in detail in Wickman and Grolemund (2017), with its own website at https://www.tidyverse.org/. He has also written an advanced book on programming R, now in its second edition (Wickman (2019)). You will find most of his work available online; follow the links in the list of references at the end of this tutorial.
The official R manuals are available as PDF files that come with the R distribution. These include An Introduction to R (a nice 100-page introduction), a manual on R Data Import/Export describing facilities for transferring data to and from other packages, and useful notes on R installation and Administration. More specialized documents include a draft of the R Language Definition, a guide to Writing R Extensions, documentation on R Internals including coding standards, and finally the massive R Reference Index (~3000 pages). The online help facility is excellent. For additional references see the annotated list at R Books.
Continue with Getting Started