Re: curve fit-in scatter, not regression.

I had several off-line exchanges with Yi Wang and here is the
final communication:

It appears that there is no NCL function that will do a least-squares
quadratic fit to scattered data. If you are conversant with
matrix methods, you can apply the singular values obtained
from a singular value decomposition to obtain the coefficients
of a polynomial fit. You can get an SVD using:

http://www.ncl.ucar.edu/Document/Functions/Built-in/svd_lapack.shtml

If a spline fit will do, try using ftcurv. There are a couple
of examples at:

http://www.ncl.ucar.edu/Document/Functions/Built-in/ftcurv.shtml

Or try ftcurvs:

http://www.ncl.ucar.edu/Document/Functions/Built-in/ftcurvs.shtml

We have put it on our list to provide a higher-level least-squares
fit function.

[Breaking news] Dennis Shea has dusted off an old copy of
FSPOLY that we should be able to get into the package soon.

Fred Clare

On May 24, 2006, at 12:20 PM, Yi Wang wrote:

> Hello NCLers:
>
> I cannot find any examples for plotting a curve to fit in the
> scatter plotting. Note I do not want a regression line, I want a curve
> to fit in my scatter plotting to capture the major patter there. Any
> help, suggestion will be welcome. Thanks in advance.
>
> Yi Wang
> _______________________________________________
> ncl-talk mailing list
> ncl-talk_at_ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk

_______________________________________________
ncl-talk mailing list
ncl-talk_at_ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Thu May 25 2006 - 12:20:24 MDT