The Origin Forum
File Exchange
Try Origin for Free
The Origin Forum
Home | Profile | Register | Active Topics | Members | Search | FAQ | Send File to Tech support
Username:
Password:
Save Password
Forgot your Password? | Admin Options

 All Forums
 Origin Forum
 Origin Forum
 Model comparison and Number of Parameters
 New Topic  Reply to Topic
 Printer Friendly
Author Previous Topic Topic Next Topic Lock Topic Edit Topic Delete Topic New Topic Reply to Topic

SapereAude1490

Slovenia
4 Posts

Posted - 06/24/2022 :  07:24:20 AM  Show Profile  Edit Topic  Reply with Quote  View user's IP address  Delete Topic
OriginPro 2018(64-bit)
Operating System: Windows 10

I have a question regarding the Fit Comparison of Models - the Bayesian and Akaike's Information Criterion Test.

I have a data set from which I'm deconvolving two or three lognormal distributions.

When I do the deconvolution and then compare the two models, I get the following:



Under the "Number of Parameters", why does it say 2 and 3 for the two compared models? Shouldn't it be 4 and 6 (2 mu's and sigma's, or 3 mu's and sigmas)? I'm a bit confused how does model comparison work with deconvolutions.

The two deconvolutions are shown below:



If anyone can explain how origin is doing this, I would be very grateful.

SapereAude1490

Slovenia
4 Posts

Posted - 06/24/2022 :  08:56:46 AM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
So after digging a bit more, I found that origin is doing this:



But, if I try to do it manually (in MATLAB), I get different results for the AIC:



Why is there a difference between the AIC results in origin and matlab? Is it a rounding thing?
Go to Top of Page

Sam Fang

281 Posts

Posted - 06/27/2022 :  02:29:17 AM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Hi SapereAude1490,

Did you fix parameters in your fitting? Note that fixed parameters are not counted in K.

K is not the number of parameters, but the number of parameters plus 1. In your matlab example K=3.



Sam
OriginLab Technical Services
Go to Top of Page

SapereAude1490

Slovenia
4 Posts

Posted - 06/27/2022 :  04:42:12 AM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Hello Sam,

Thanks for your response. I didn't know K was number of parameters + 1. Thanks.

I tried doing the deconvolution again and not fixing any of the parameters (I did give it custom starting positions).

I got the exact same result as before. Is it working correctly with reporting of 2 and 3 parameters?



I also tried to change the K to K+1 in matlab and it does produce the same results now.

Go to Top of Page

Sam Fang

281 Posts

Posted - 06/27/2022 :  05:29:02 AM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Hi SapereAude1490,

I can't reproduce your problem for number of parameters in Origin.

Can you send your project file to us? We can check it.

Thanks.

Sam
OriginLab Technical Services
Go to Top of Page

SapereAude1490

Slovenia
4 Posts

Posted - 06/27/2022 :  05:48:25 AM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Dear Sam,

Thanks for the quick reply. Here's the file:

https://my.originlab.com/ftp/forum_and_kbase/Images/DeconvolutionProblem3vs2.opju
Go to Top of Page

Sam Fang

281 Posts

Posted - 06/27/2022 :  10:46:58 PM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Hi SapereAude1490,

I still can't reproduce your result.

I noticed that in the Fit Comparison report, Fit Reports are [Data]FitPeaks2x and [Data]FitPeaks3x. However [Data] book has no such sheets. So the comparison report may not use fit reports in current project?

Sam
OriginLab Technical Services
Go to Top of Page
  Previous Topic Topic Next Topic Lock Topic Edit Topic Delete Topic New Topic Reply to Topic
 New Topic  Reply to Topic
 Printer Friendly
Jump To:
The Origin Forum © 2020 Originlab Corporation Go To Top Of Page
Snitz Forums 2000