Prev: NARX for prediction of WWTP effluent flow?
Next: xpc target rs232 send problem
From: Valeria Fabbri fabbri on 17 Feb 2010 10:17
Hi,
I have a problem to choose the method to use on the griddata function. Which method is better? the v4 method or the cubic method? What are the differences between them? I have few real data (about 14) and a large number of grid point (about 50).
Have you got a bibliography for the v4 method?
Than you very much.
From: John D'Errico on 17 Feb 2010 11:27
"Valeria Fabbri fabbri" <valeriafabbri.fisica(a)gmail.com> wrote in message <hlh1a4$ts$1(a)fred.mathworks.com>...
> Hi,
> I have a problem to choose the method to use on the griddata function. Which method is better? the v4 method or the cubic method? What are the differences between them? I have few real data (about 14) and a large number of grid point (about 50).
> Have you got a bibliography for the v4 method?
> Than you very much.

I seem to recall the 'v4' method is a variation of
radial basis function interpolant. If you want to
learn more, there will be a reference in the code,
or in

doc griddata

You can also do a google search for "radial basis
function interpolation".

This method will tend to be terribly poor for large
set of data, as it forces the solution of a (full)
system of linear equations as large as the number
of data points. The v4 method will also be poor
in extrapolation, but at least it CAN extrapolate.

The cubic method is a triangulation based method,
so it cannot extrapolate, and it may do strange
things near the boundaries of the domain. But it
will be more efficient for larger sets of data. You
don't have a large set of data.

You are interpolating based on only 14 data points
in two dimensions? Sigh. Don't expect any method
to give very good estimates.

John
 | 
Pages: 1
Prev: NARX for prediction of WWTP effluent flow?
Next: xpc target rs232 send problem