Quantcast

Question on Using Experimenter for Text Classification problems

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Question on Using Experimenter for Text Classification problems

SanjayPethe
Hello,
I was trying to use the Experimenter to compare classifiers for text classification and had the following questions:
  1. I tried using a FilteredClassifier with the ClassAssigner, NominalToString and StringToWordVector filters in a MultiFilter. However, the Experimenter did not like the class being reassigned, and it appears that the Experimenter requires the Class attribute to be the last attribute. Since the StringToWord filter always changes this, it appears that this filter cannot be used in the Experimenter. Is this correct?
  2. I was able to compare classifiers for this purpose using preprocessed files with these filters already applied and the class element moved to be the last attribute. Is this the only way to use the Experimenter for this purpose?
 
Regards,
Sanjay Pethe
 
 

_______________________________________________
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
|  
Report Content as Inappropriate

Re: Question on Using Experimenter for Text Classification problems

Eibe Frank-2
Administrator
You don't need to use the ClassAssigner. StringToWordVector will change the index of the class attribute in the Instances object it generates. Subsequent classifiers in the chain will thus process the data correctly.

Cheers,
Eibe

On 17/02/2017, at 6:04 PM, Pethe, Sanjay <[hidden email]> wrote:

Hello,
I was trying to use the Experimenter to compare classifiers for text classification and had the following questions:
  1. I tried using a FilteredClassifier with the ClassAssigner, NominalToString and StringToWordVector filters in a MultiFilter. However, the Experimenter did not like the class being reassigned, and it appears that the Experimenter requires the Class attribute to be the last attribute. Since the StringToWord filter always changes this, it appears that this filter cannot be used in the Experimenter. Is this correct?
  2. I was able to compare classifiers for this purpose using preprocessed files with these filters already applied and the class element moved to be the last attribute. Is this the only way to use the Experimenter for this purpose?
 
Regards,
Sanjay Pethe
 
 
_______________________________________________
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
|  
Report Content as Inappropriate

Re: Question on Using Experimenter for Text Classification problems

SanjayPethe
In reply to this post by SanjayPethe
Thanks Eibe. Will give that a try.
 
Regards,
Sanjay Pethe
Lead, Strategic Sourcing
Work: 281-293-4495
Cell: 281-907-2443
e-mail: [hidden email]
 
 
_____________________________________________
From: Pethe, Sanjay
Sent: Thursday, February 16, 2017 11:05 PM
To: '[hidden email]' <[hidden email]>
Subject: Question on Using Experimenter for Text Classification problems
 
 
Hello,
I was trying to use the Experimenter to compare classifiers for text classification and had the following questions:
  1. I tried using a FilteredClassifier with the ClassAssigner, NominalToString and StringToWordVector filters in a MultiFilter. However, the Experimenter did not like the class being reassigned, and it appears that the Experimenter requires the Class attribute to be the last attribute. Since the StringToWord filter always changes this, it appears that this filter cannot be used in the Experimenter. Is this correct?
  2. I was able to compare classifiers for this purpose using preprocessed files with these filters already applied and the class element moved to be the last attribute. Is this the only way to use the Experimenter for this purpose?
 
Regards,
Sanjay Pethe
 
 

_______________________________________________
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
Loading...