[despammed] RE: [R] multidimensional function fitting
simon at stats.gla.ac.uk
Mon Mar 3 20:43:02 CET 2003
> 8-) > 2) It is very memory hungry, esp. when using the s() function: I have
> 8-) > 192Mb with 256Mb swap (not a lot, but reasonable I'd say), and I've
> 8-) > never had to kill R as often as when trying gam()...
> 8-) >
> 8-) - do you have a very large number of data? The way mgcv works it first
> 8-) finds an "optimal" basis for smoothing and this will involve formation of
> Some 10000 (x,y,z). For R, that doesn't strike me as particularly much.
- The point here is that the method being used by default requires storage
of a 10000 by 10000 matrix - it's not really an issue of whether 10000 is
particularly much "for R". There is a cut and paste-able example for
getting around this at the end of the help file for gam()... gam will be
*much* faster and *much* less memory intensive if you use it.
More information about the R-help