Question concerning the CFS attribute selection

[Sebastian Briesemeister]

Hello everyone,

I wondered if in case the locallyPredictive option of CFSSubsetEval is
turned off, will the class label still be considered in the evaluation???
Or will it just consider the correlation between two features without
knowing the actual class?

Is there another way in weka to exclude highly correlated features?

Thanks in advance,

Cheers,
Sebastian


_______________________________________________
Wekalist mailing list
Wekalist@...
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist

[Mark Hall-9]

On 28/06/2008, at 3:05 AM, Sebastian Briesemeister wrote:

> Hello everyone,
>
> I wondered if in case the locallyPredictive option of CFSSubsetEval  
> is turned off, will the class label still be considered in the  
> evaluation???
> Or will it just consider the correlation between two features  
> without knowing the actual class?

CFS is a supervised attribute selection method. It has to consider  
the class.

Cheers,
Mark.

--
Mark Hall
Senior Developer/Consultant, Pentaho Open Source Business Intelligence
Citadel International, Suite 340, 5950 Hazeltine National Dr.,
Orlando, FL 32822, USA
[phone] office, [phone] mobile, [phone] fax,
Skype: mark.andrew.hall, Yahoo: mark_andrew_hall
Download the latest release today <http://www.sourceforge.net/ 
projects/pentaho>




_______________________________________________
Wekalist mailing list
Wekalist@...
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist