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
 Clustering analysis
 New Topic  Reply to Topic
 Printer Friendly
Author Previous Topic Topic Next Topic Lock Topic Edit Topic Delete Topic New Topic Reply to Topic

Neena1695

Canada
83 Posts

Posted - 06/30/2022 :  10:17:40 PM  Show Profile  Edit Topic  Reply with Quote  View user's IP address  Delete Topic
Origin Ver: 2022
Operating System: Windows 10

Hello Origin users,

Is there a way on Origin to do a clustering analysis on this data, so to separate this dataset into 2 different clusters (marked in circles in the below picture)?

Thank you,
Neena

Shirley_GZ

China
Posts

Posted - 07/01/2022 :  01:27:17 AM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Hi Neena,

Please try to use the "Cluster" gadget to set different groups,
https://www.originlab.com/doc/Origin-Help/Gadget-Cluster

And then, use this app to draw ellipses,
https://www.originlab.com/FileExchange/details.aspx?fid=365

Thanks,
Shirley

Originlab Technical Service Team

Edited by - Shirley_GZ on 07/01/2022 01:48:08 AM
Go to Top of Page

aplotnikov

Germany
109 Posts

Posted - 07/01/2022 :  03:41:07 AM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
I am not sure that clustering is suitable for this purpose at all. There are special methods to detect abrupt signal changes, e.g.:
Detection of Abrupt Changes - Theory and Application
by Michèle Basseville and Igor V. Nikiforov,
Prentice-Hall, 1993
ISBN 0-13-126780-9
http://people.irisa.fr/Michele.Basseville/kniga/kniga.pdf
Go to Top of Page

Neena1695

Canada
83 Posts

Posted - 07/05/2022 :  11:35:47 AM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
I realize that clustering cannot be used; is there a way on Origin to split my Y values in column into two clusters based on specified intervals say 260-270 and 270-280? Categorize it as say 1 and 2 respectively? Then I can color map using that category.
Go to Top of Page

snowli

USA
1080 Posts

Posted - 07/05/2022 :  12:00:40 PM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Please try Worksheet: Split Worksheet tool
https://www.originlab.com/doc/Origin-Help/split-wks-into-sheets

When split mode is by reference column, there are several ways to plit
By direction change, direction reset, by value change.

Please see if one of them is what you meant by "abrupt signal changes"

Thanks, Snow



quote:
Originally posted by aplotnikov

I am not sure that clustering is suitable for this purpose at all. There are special methods to detect abrupt signal changes, e.g.:
Detection of Abrupt Changes - Theory and Application
by Michèle Basseville and Igor V. Nikiforov,
Prentice-Hall, 1993
ISBN 0-13-126780-9
http://people.irisa.fr/Michele.Basseville/kniga/kniga.pdf

Go to Top of Page

snowli

USA
1080 Posts

Posted - 07/05/2022 :  12:02:51 PM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Cluster gadget can color subrange of plot. See suggestion 3 on https://www.originlab.com/doc/Quick-Help/color-datapoints-regions

Thanks, Snow

quote:
Originally posted by Neena1695

I realize that clustering cannot be used; is there a way on Origin to split my Y values in column into two clusters based on specified intervals say 260-270 and 270-280? Categorize it as say 1 and 2 respectively? Then I can color map using that category.

Go to Top of Page

aplotnikov

Germany
109 Posts

Posted - 07/05/2022 :  12:52:41 PM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
quote:
Originally posted by snowli

Please try Worksheet: Split Worksheet tool
https://www.originlab.com/doc/Origin-Help/split-wks-into-sheets

When split mode is by reference column, there are several ways to plit
By direction change, direction reset, by value change.

Please see if one of them is what you meant by "abrupt signal changes"


Dear snow,

what I meant by "abrupt signal changes" is comprehensively described in the book I referred to.
The problem has nothing to do with clustering - you just need to recognize statistically significant changes of a very noisy signal. The simplest way is to smooth data (e.g., by median filtering) and check whether the smoothed values are out of the range of variation (given or calculated from signal statistics). But indeed the problem is a bit more complicated - you can find details in the suggested book. The splitting is very simple if you know exactly the criterium. "Value change" for noisy signals is rather not so evident and frequently requires certain tests of statistic hypotheses.

Regards,

Alexei
Go to Top of Page

Neena1695

Canada
83 Posts

Posted - 07/05/2022 :  3:56:30 PM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Hi Shirley, Snow and Alexei,

I just want to do a simple splitting of the data into categories based on value with a larger tolerance (10 units) than what currently split worksheet does.
For example, my lowest value in the column is 260 and highest is 280. How to split data into two categories say 260-270 as category 1, and 270-280 as category 2.

Any help would be appreciated.
Thank you,
Neena

Go to Top of Page

snowli

USA
1080 Posts

Posted - 07/05/2022 :  4:15:40 PM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Hi Neena,

Try Worksheet: Split Columns... instead.
Choose Subgroup method as By Reference Column.
And specify the column with 260, ... 280 values.
Then choose Split by Interval
Set Interval to be be 10.001 //if set it to 10, 280 will be in new column.
Start value is 260.


Thanks, Snow
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