| Author |
 Topic  |
|
|
Marco1234
Germany
7 Posts |
 Posted - 06/18/2002 : 03:38:36 AM
|
Hello,
I have to fit a function like y=C1*X1+C2*X2...C24*X24 with Data1_a, Data1_b...Data1_x as depandable Var and Data2_b as independable Var!
My Athlon takes about 20 Minutes for one fit! Can I speed up this procedure? Or should I use other commands?
(I am using:
nlsf.func$
nlsf.y$
nlsf.x$
nlsf.iterate(10)
nlsf.funccol$
nlsf.makecurve(func))
I tried tp write this in Origin C but it was not faster, because I have also to use the "iterate" procedure!
More than 1000 Datasets must be fitted but with my routine that would take weeks over weeks! Can someone help me? |
|
|
easwar
USA
1965 Posts |
Posted - 06/18/2002 : 08:56:44 AM
|
Hi,
If you create a new function and set this function to use Origin C (by checking the compile check box in Edit page of NLSF), and then save that function and use it for fitting in your LabTalk script, it should work much faster than you report.
In any case, it looks like what you want is just multiple linear regression. Origin's built-in multiple regression tool restricts the number of independant variables to 9. However, if you have ver 7, you can use the NAG library to perform multiple regression for more than 9 independant variables. Look up the NAG function nag_regsn_mult_linear for further documentation. Using the NAG routine should work out to be much faster for you.
Easwar
OriginLab.
Edited by - easwar on 06/18/2002 08:57:45 |
 |
|
|
Marco1234
Germany
7 Posts |
Posted - 06/19/2002 : 03:20:13 AM
|
Thank you, that helps a lot!
I will use the "Compile before use" feature!
Now it takes about 5 Seconds time!
It is a wonderful feature in Origin 7!
bye
|
|
|
|
JoannaZajac
United Kingdom
14 Posts |
Posted - 10/26/2011 : 3:48:42 PM
|
Hi, is there a similar feature for orign 6?
I use fit in a loop for datasets which have 1000 columns, fit for each, it takes 5 mins to go through dataset, which is very long as I have hundreds of datasets,
appriciate comments!!
Joanna |
|
|
| |
Topic |
|
Speed up NLSF