Re: lspoly failed

lspoly does do a poor job. Not sure why. For only a 2nd order
polynomial, the approach used by lspoly should be fine.

I tried a simple scaling of x and y (*0.1). See the attached file.
It worked fine. Hence, there must have been a numric
issu. Again, not sure why.

I also compared the result to a simple linear regression.
There is essentially no difference in the lines.
See print values.

see attached

On 11/25/13 8:48 PM, wuhui wrote:
> Hi folks,
>
>
> I followed scatter_9.ncl in
> http://www.ncl.ucar.edu/Applications/interp1d.shtml to fit a
> curve. But when the input data (e.g., x, y) was replaced by my
> series, the fitted curve is completely out of data range, which
> is obviously wrong.
>
> The script, data and output figure are attached.
>
> A trivial slip is coefficients should be timed by "x" rather
> than "y" in line 24 of original example script. Am I right?
>
>
> Hui Wu
>
>
>
> _______________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk