SuperGEM(tm) is an implementation of the patient rule induction method (PRIM) for optimizing noisy functions of many variables described in Bump Hunting in High-Dimensional Data (compressed PostScript).
SuperGEM(tm) 1.0 provides a rudimentary interface for using PRIM with S [Becker, Chambers and Wilks(1988)] or Splus(r) [ Statistical Sciences, Inc., Seattle, WA].
It is assumed that S or Splus has already been installed on the computer. Details of the SuperGEM installation procedure depend upon the host platform upon which S/Splus in running. The two possibilities are UNIX, and a PC compatible computer running a Microsoft operating system (WINTEL). The first step for both is to choose a directory in which to store the files associated with SuperGEM. The full path name of this directory will be referred to as dir. Examples might be /home/your_name/gem (UNIX), or c:\gem (WINTEL).
Here dir is the directory where the SuperGEM files have been stored.
For these commands substitute / for each \ in the definition of dir.
After installation enter
>gemhelp()
at the S/Splus command line for information concerning documentation of the SuperGEM / S interface.
The PRIM procedure is described in Friedman and Fisher (1997) (compressed PostScript).
Becker, R. A., Chambers, J.M., and Wilks, A. R. (1988). The New S Language. Wadsworth & Brooks/Cole.
Friedman, J. H. and Fisher, N. I. (1997). Bump hunting in high dimensional data (compressed PostScript).
www@stat.stanford.edu