Stat500 resources

R and S+ Resources

The statistical software I am using for the course is R. You can download copies of the software and get other information from the R Home Page.

The most recent version of R for Windows is version 0.651. See the R on Windows FAQ for help on the installation and see some additional tips on using R in Windows

R is based on the ( award winning) S language developed at Bell Labs by John Chambers. The S language is described in the The New S Language by Becker, Chambers & Wilks and Statistical Models in S by Chambers and Hastie, both published by CRC press. Almost all the commands described in these two books are implemented in R. The recommended text for the course is Modern Applied Statistics with S-PLUS by W. Venables and B. Ripley (3nd Edition). The authors have written some online complements which (among others) describe the differences between R and S-PLUS.

S+ is a commercial software package based on S distributed by Mathsoft. S+ is available on Statistics Department computers, the ITD Statistical Computing servers and other locations on Campus. Almost all the commands in the course text will work in S+ also. Sometimes minor changes will be necessary and the output may differ slightly. S+ for Windows comes with a GUI interface which is easier to get started with but you won't be able exercise the full power of the language without knowing the command line versions.

Most of the materials written for S+ apply equally well to R:

Univ of Waterloo guide to S+
Notes on S+ by Bill Venables This is an old text from which the recommended text was developed. The book is preferred.
S+ tutorial from Dalhousie Univ
S+ information from ASU

You can find additional information about Splus on the department's S+ page.

R packages

The following R packages are used in Stat 500. We won't be needing these for a while so you can postpone installing them if you wish.

MASS - this comes as part of the VR package that goes with Venables and Ripleys book.
leaps - for variable selection. If you are using S+, you won't need to install this because it is part of the base S+ package.
splines - for computing splines
mva - for multivariate analysis. This comes with the basic R installation so shouldn't need to specially install this.

In addition you will need to get some functions for drawing ellipses. There is an S+ library available from STATLIB's S archive which will run without modification under R. I have repackaged this S+ library for use under R ( zip file for Windows and tarball for Unix). I have not tested all the functions or ported the documentation but it does perform as advertized for our purposes.

In addition, I have compiled own package which includes all the additional functions that I wrote for use in the course along with the datasets. Available as a zip file for Windows and tarball for Unix). I have not documented the functions or data yet.

Julian Faraway - <faraway@umich.edu>

Last modified on 10/13/99