Prev: Need help desperately: lsqlin
Next: accumarray, histc, bin in non-monotonic bins?
From: Tay on 27 May 2010 11:00
Hi all,

I have to fit a 3D curve wich has as X and Y values two pressure coefficients (they have been calculated from a probe calibration). For every value of X and Y corresponds a Value Z wich is an angle.

Now what I need is a curve fit wich gives me a formula so that giving the two coefficient X and Y as input i will get the corresponding angle.

Can anyone help me?
Thanks
From: Steven Lord on 27 May 2010 12:45

"Tay " <taytus(a)hotmail.com> wrote in message
news:htm1e7$mo9$1(a)fred.mathworks.com...
> Hi all,
>
> I have to fit a 3D curve wich has as X and Y values two pressure
> coefficients (they have been calculated from a probe calibration). For
> every value of X and Y corresponds a Value Z wich is an angle.
>
> Now what I need is a curve fit wich gives me a formula so that giving the
> two coefficient X and Y as input i will get the corresponding angle.
>
> Can anyone help me?

http://www.mathworks.com/support/solutions/en/data/1-16JQD/index.html?solution=1-16JQD&BB=1

--
Steve Lord
slord(a)mathworks.com
comp.soft-sys.matlab (CSSM) FAQ: http://matlabwiki.mathworks.com/MATLAB_FAQ
To contact Technical Support use the Contact Us link on
http://www.mathworks.com


From: Walter Roberson on 27 May 2010 13:59
Tay wrote:

> I have to fit a 3D curve wich has as X and Y values two pressure
> coefficients (they have been calculated from a probe calibration). For
> every value of X and Y corresponds a Value Z wich is an angle.
>
> Now what I need is a curve fit wich gives me a formula so that giving
> the two coefficient X and Y as input i will get the corresponding angle.

There is not (and cannot be) a unique or meaningful answer unless you
already know the form the formula should take. Think of your known X, Y,
and Z as fixed physical points and think of there being elastic sheets
connecting all of the points to their neighbours: you can pinch or
stretch any of the elastic sheets an infinite number of ways, and each
different way represents a different possible formula that completely
explains the data. If you do not have some knowledge ahead of time that
constrains the possibilities, you have no way of saying that any of the
infinite number of formula are more correct than any of the others, and
the probability that you will hit upon the correct formula would be 0.
From: zunzun on 16 Jun 2010 16:36
Try the online 3D "Function Finder" at zunzun.com, the link is

http://zunzun.com/FunctionFinder/3/

Set the smoothness control to use only equations with two or three
coefficients as first, as a simpler function will generally
interpolate and extrapolate more smoothly than a complex one. Repeat
the procedure with increasing number of coefficients until you find a
suitable function. Note that turning on the "Polyfunctional" family
of equations will give large numbers of basically random functions,
you should probably turn these off at first so you have known named
equations only.

James Phillips
http://zunzun.com
zunzun(a)zunzun.com

On May 27, 10:00 am, "Tay " <tay...(a)hotmail.com> wrote:
>
> Now what I need is a curve fit wich gives me a formula so that giving the two coefficient X and Y as input i will get the corresponding angle.
 | 
Pages: 1
Prev: Need help desperately: lsqlin
Next: accumarray, histc, bin in non-monotonic bins?