Readme file

SERIES B
Statistical Methodology

Modelling and smoothing parameter estimation with multiple quadratic penalties, by S. N. Wood
Journal of the Royal Statistical Society Series B, Statistics in Society, Volume 62 (2000).

The accompanying file contains code implementing the multiple smoothing parameter estimation method described in the paper. The method is supplied as an R package (http://www.ci.tuwien.ac.at/R/ )  which has been  zipped (see http://www.cdrom.com/pub/infozip/   for free (un)zipping software).

Once unzipped, the directory mgcv is a complete R package, and can be loaded directly using the windows version of R. For Linux/Unix unpack mgcv.tar (also in the zip file) using tar, and install in the usual way with a command like `R INSTALL mgcv'.

Within R the package is loaded with something like: library(mgcv) The routines mgcv(), GAMsetup(), QT() and SANtest() are available along with R documentation for each.

The code also works with S-PLUS - see the subdirectories mgcv/src for C source code, and mgcv/R for S routines. Further documentation for the C source code and the R/S routines can be found in: mgcv/man/mgcv.tex|.ps|.pdf.

Note that the C routines can be used without R or S-PLUS, in which case only the .c and .h files in mgcv/src are needed along with the documentation in mgcv/man/mgcv.ps (or .tex or .pdf).

For possible updates see:
http://www.ruwpa.st-and.ac.uk/simon.html

Please send comments and bug reports to snw@st-and.ac.uk

Simon Wood
Mathematical Institute
North Haugh,
St Andrews
Fife
KY16 9SS
UK

Journals

SERIES A
Statistics in Society

SERIES B
Statistical Methodology

SERIES C
Applied Statistics

SERIES D
The Statistician