s-news
[Top] [All Lists]

FW: [S] ppreg in splus

To: "'ripley@stats.ox.ac.uk'" <ripley@stats.ox.ac.uk>, "'s-news@wubios.wustl.edu'" <s-news@wubios.wustl.edu>
Subject: FW: [S] ppreg in splus
From: "Kurbat, Matthew A." <matthew.a.kurbat@ncmi.com>
Date: Tue, 19 Oct 1999 10:36:57 -0400
Cc: "'jon.frye@chi.frb.org'" <jon.frye@chi.frb.org>, "'kurbatm@hotmail.com'" <kurbatm@hotmail.com>
Sender: owner-s-news@wubios.wustl.edu

Thank you Professor Ripley!

> -----Original Message-----
> From: Prof Brian D Ripley [SMTP:ripley@stats.ox.ac.uk]
> Sent: Tuesday, October 19, 1999 2:13 AM
> To:   Kurbat, Matthew A.
> Subject:      Re: [S] ppreg in splus
> 
> On Mon, 18 Oct 1999, Kurbat, Matthew A. wrote:
> 
> > 
> > 
> > 
> > I have a ppreg (projection pursuit regression) question.  
> > 
> > ppreg has the following optional argument
> >     "optlevel=      
> > integer from 0 to 3 which determines the throughness of an optimization
> > routine in ppreg. A higher number means more optimization."
> > 
> > Does anyone know in more detail exactly what this means?
> 
> Yes. You need to read:
> 
>        Friedman, J. H. (1984) SMART User's Guide.  Laboratory for
>        Computational  Statistics,  Stanford  University Technical
>        Report No. 1.
> 
> since that is where the code comes from. I only have a photocopy of it.
> Alternatively, look at the help pages for my ppr (library MASS):
> 
>        The basic method is  given  by  Friedman  (1984),  and  is
>        essentially  the  same  code  used  by ppreg.  The code is
>        extremely sensitive to the compiler used, and this version
>        uses double precision to try to reduce differences between
>        machines.  The differences are the ability to  use  spline
>        smoothers and the interface which should be much easier to
>        use.
> 
>        The algorithm first adds up to max.terms ridge  terms  one
>        at a time; it will use less if it is unable to find a term
>        to add that makes sufficient difference.  It then  removes
>        the  least "important" term at each step until nterm terms
>        are left.   The  levels  of  optimization  differ  in  how
>        thoroughly  the  models  are refitted during this process.
>        At level 0 the existing ridge terms are not  refitted.  At
>        level  1  the  projection directions are not refitted, but
>        the ridge functions and the regression  coefficients  are.
>        Levels  2 and 3 refit all the terms and are equivalent for
>        one response; level 3 is more careful  to  re-balance  the
>        contributions from each regressor at each step and so is a
>        little less likely to converge to a saddle  point  of  the
>        sum of squares criterion.
> 
> 
> -- 
> Brian D. Ripley,                  ripley@stats.ox.ac.uk
> Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
> University of Oxford,             Tel:  +44 1865 272861 (self)
> 1 South Parks Road,                     +44 1865 272860 (secr)
> Oxford OX1 3TG, UK                Fax:  +44 1865 272595
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu.  To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message:  unsubscribe s-news

<Prev in Thread] Current Thread [Next in Thread>