PCA issues when trying to compute it manually

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

PCA issues when trying to compute it manually

David Campa
Hi,

I'm an Audiovisual Systems Engineering student and i'm using Weka PCA as a
part of my final project degree.

I have noticed that if I apply weka's PCA filter to a dataset I get a result
but, if extract eigenvectors and then I multiply the original dataset by its
eigenvectors I don't get the same result. In fact, the results are very
different.

I don't know why I'm getting this results because I should get the same,
right?

Thank you for your support,

David.





--
Sent from: http://weka.8497.n7.nabble.com/
_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: PCA issues when trying to compute it manually

Martin
Use PCA in conjunction with FilteredClassifier. This is the reliable way.

Regards, 
Martin 

On 27 Feb 2018 8:56 a.m., "David Campa" <[hidden email]> wrote:
Hi,

I'm an Audiovisual Systems Engineering student and i'm using Weka PCA as a
part of my final project degree.

I have noticed that if I apply weka's PCA filter to a dataset I get a result
but, if extract eigenvectors and then I multiply the original dataset by its
eigenvectors I don't get the same result. In fact, the results are very
different.

I don't know why I'm getting this results because I should get the same,
right?

Thank you for your support,

David.





--
Sent from: http://weka.8497.n7.nabble.com/
_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html



If you reply to this email, your message will be added to the discussion below:
http://weka.8497.n7.nabble.com/PCA-issues-when-trying-to-compute-it-manually-tp42712.html
To unsubscribe from WEKA, click here.
NAML

_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: PCA issues when trying to compute it manually

Eibe Frank-3
In reply to this post by David Campa
Did you apply standardization to your data? WEKA's PrincipalComponents standardizes the data (by default) before performing PCA. (You can configure it to only center the data instead.) PrincipalComponents also performs standardization (or centering) before filtering an instance by multiplying it with the eigenvectors.

Cheers,
Eibe

On Tue, Feb 27, 2018 at 1:49 PM, David Campa <[hidden email]> wrote:
Hi,

I'm an Audiovisual Systems Engineering student and i'm using Weka PCA as a
part of my final project degree.

I have noticed that if I apply weka's PCA filter to a dataset I get a result
but, if extract eigenvectors and then I multiply the original dataset by its
eigenvectors I don't get the same result. In fact, the results are very
different.

I don't know why I'm getting this results because I should get the same,
right?

Thank you for your support,

David.





--
Sent from: http://weka.8497.n7.nabble.com/
_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html


_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: PCA issues when trying to compute it manually

David Campa
Hi Eibe,

Applying standardization to the original data was the key! 

Now, WEKA's PCA and applying the original data before standardizing * eigenvectors gives me the same result.

Thanks a lot for your support!

David

2018-02-28 8:40 GMT+01:00 Eibe Frank <[hidden email]>:
Did you apply standardization to your data? WEKA's PrincipalComponents standardizes the data (by default) before performing PCA. (You can configure it to only center the data instead.) PrincipalComponents also performs standardization (or centering) before filtering an instance by multiplying it with the eigenvectors.

Cheers,
Eibe

On Tue, Feb 27, 2018 at 1:49 PM, David Campa <[hidden email]> wrote:
Hi,

I'm an Audiovisual Systems Engineering student and i'm using Weka PCA as a
part of my final project degree.

I have noticed that if I apply weka's PCA filter to a dataset I get a result
but, if extract eigenvectors and then I multiply the original dataset by its
eigenvectors I don't get the same result. In fact, the results are very
different.

I don't know why I'm getting this results because I should get the same,
right?

Thank you for your support,

David.





--
Sent from: http://weka.8497.n7.nabble.com/
_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html


_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html



_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html