| Author |
 Topic  |
|
|
antonio2
Brazil
Posts |
 Posted - 01/13/2006 : 2:53:48 PM
|
Origin Version (Select Help-->About Origin): 7.5
Operating System:win xp
What is the maximum degree of freedom on Multiple linear regression. The sample shows 3, but I need more than ten, which way can we extend it. ?
|
|
|
easwar
USA
1965 Posts |
Posted - 01/13/2006 : 3:09:44 PM
|
Hi,
The current limit is 9 independent variables. We will address this limitation in the next version.
For now you have two options. One is to see if the NAG library has suitable functions and write custom Origin C code to access those functions to perform the regression.
The second option is to use the NLSF tool and do minimization of a user-defined function. This may be easier in that it is accessible from GUI, but perhaps a bit quirky because you have to edit your function to add another term etc.
Say you want 10 indep vars. With NLSF what you can do is:
1> define a new function
2> set indep var names as
xx1,xx2,xx3,xx4,xx5,xx6,xx7,xx8,xx9,xx10
3> set dep var name as
y
4> Set user defined parameter names as:
a0,a1,a2,a3,a4,a5,a6,a7,a8,a9,a10
5> Define the function as:
y=a0+a1*xx1+a2*xx2+a3*xx3+a4*xx4+a5*xx5+a6*xx6+a7*xx7+a8*xx8+a9*xx9+a10*xx10
6> Assign all indep columns (typically cols 2 thru 11 for the 10-indep var case) to the xx vars. Assign dep col (typically 1st col) for the y var
7> Simply assign value = 1 for all parameters and do the iteration and let the fitter find the optimal parameter values and their associated errors.
Easwar
OriginLab
Edited by - easwar on 01/13/2006 3:29:56 PM |
 |
|
| |
Topic |
|
|
|
multiple linear regression what is the max