| Re: Help: multiple linear regression fit [message #36185] |
Tue, 12 August 2003 19:54 |
fishdick91
Messages: 2
Registered: August 2003
|
Junior Member |
|
|
Craig Markwardt <craigmnet@cow.physics.wisc.edu> wrote in message news:<onoeyurhtw.fsf@cow.physics.wisc.edu>...
> And what of the MPFIT family of functions? Using the driver
> MPFITEXPR, you can set up an expression that almost exactly matches
> Some people seem to resist using a non-linear regression tool for a
Yes, we once thought so, although my colleagues and I have used your
tools to fit all kinds of spectra lines. :)
Now it seems the time for us to change our mind, and use MPFITEXPR to
do 'linear regression' also.
Thank you, for your procedures and your advice.
Good luck,
Dick
> linear problem on the basis of, "it's *ONLY* linear regression!"
> Originally I might explain that the heart of any non-linear regression
> package is a linear solver, so it's really all the same. Recently I
> just shrug and say, "their loss."
>
> Happy fitting,
> Craig
>
> http://cow.physics.wisc.edu/~craigm/idl/idl.html (under fitting)
|
|
|
|
| Re: Help: multiple linear regression fit [message #36195 is a reply to message #36185] |
Tue, 12 August 2003 08:55  |
Craig Markwardt
Messages: 1869
Registered: November 1996
|
Senior Member |
|
|
fishdick91@hotmail.com (fishdick91@hotmail.com) writes:
> Hi everyone,
>
> I've set up a model: y = a1*x1 + a2*x2 + a3*x3 to fit my experiment
> data,
> and am using 'regress' function to perform this fit.
> However, the 'regress' function always return a big const which I
> don't need.
> So question 1:
> how can I fix the const when fitting?
And what of the MPFIT family of functions? Using the driver
MPFITEXPR, you can set up an expression that almost exactly matches
your case:
EXPR = 'P(0)*X(*,0) + P(1)*X(*,1) + P(2)*X(*,2)'
(of course you need the core routine MPFIT too). You would need to
set up X as an Nx3 array, and Y as an N-vector.
Some people seem to resist using a non-linear regression tool for a
linear problem on the basis of, "it's *ONLY* linear regression!"
Originally I might explain that the heart of any non-linear regression
package is a linear solver, so it's really all the same. Recently I
just shrug and say, "their loss."
Happy fitting,
Craig
http://cow.physics.wisc.edu/~craigm/idl/idl.html (under fitting)
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
