[R] multidimensional function fitting

[Wiener, Matthew]

Take a look at package mgcv.  Hope this helps.  --Matt

-----Original Message-----
From: RenE J.V. Bertin [mailto:rjvbertin at despammed.com]
Sent: Thursday, February 27, 2003 1:39 PM
To: r-help at stat.math.ethz.ch
Subject: [R] multidimensional function fitting


Hello,

I have been looking around for how to perform a multidimensional, arbitrary
function fit (in any case non-linear; more below), but haven't found
anything yet on how to accomplish this. I found some references to the
funfits package (that apparently is now called fields), but as far as I have
understood it, it does not do what I need.

Basically, I have a set of data that describe some measured, dependent
variable as a function of a number of controlled variables, in casu
z=f(x,y). What I need is to obtain some f that reliably describes my data on
the range of controlled variables. (This is for inverse-filtering in a
visual stimulus programme, so I don't really care about the form of f as
long as it is reasonably simple to implement -- i.e. a spline surface would
not be ideal in the context ;^). )

Clearly, something like

fit<-lm(z~x+y+x*y+x*x+y*y, data)

does not give the result I want.

Is there a package somewhere that allows to do this sort of operation? Like
a multidimensional version of polyreg()?

Thanks in advance,
RenE Bertin

______________________________________________
R-help at stat.math.ethz.ch mailing list
http://www.stat.math.ethz.ch/mailman/listinfo/r-help

------------------------------------------------------------------------------