defining attributes and classes

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

defining attributes and classes

Michael Thon
Greetings - I have a multiclass dataset that I imported into weka.  I
would like to try the multiclasslassifier with naiveBayesSimple .  It
seems to be returning only 1 classifier (only one set of probabilities)
for the class selected in the dropdown list on the left side of the
screen.  How can I select multiple classes to be trained?
Cheers
Mike


_______________________________________________
Wekalist mailing list
[hidden email]
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Reply | Threaded
Open this post in threaded view
|

Re: defining attributes and classes

Surendra Singhi
Michael Thon <[hidden email]> writes:

> Greetings - I have a multiclass dataset that I imported into weka.  I

If I understand you correctly, then you mean dataset with instances having
more than one class.

> would like to try the multiclasslassifier with naiveBayesSimple .  It
> seems to be returning only 1 classifier (only one set of probabilities)
> for the class selected in the dropdown list on the left side of the
> screen.  How can I select multiple classes to be trained?

The term multiclassclassifier as used in Weka is misleading, if you read the
documentation; it says
"Class for handling multi-class datasets with 2-class distribution
classifiers"

what it means, is that this(MCC) classifier can be used to handle datasets
with more than 2 classes by using binary classifiers(classifiers which take
only 2 classes). For example say a classification problem with 10 classes, and
you want to make use of SVM which is a binary classifier.

So, getting back to your problem, currently it is not possible to do
multi-class classification using Weka, unless you extend Weka.

Eibe, please correct me, if I am wrong.

Cheers.
--
Surendra Singhi
http://www.public.asu.edu/~sksinghi/index.htm

Great wits are sure to madness near allied,
And thin partitions do their bounds divide.

   (John Dryden, Absalom and Achitophel, 1681)


_______________________________________________
Wekalist mailing list
[hidden email]
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Reply | Threaded
Open this post in threaded view
|

Re: defining attributes and classes

Eibe Frank
Yes, you are right, but the term "multi-class dataset" is commonly used
to refer to problems with more than two class values (so the
terminology in Weka is not misleading).

If you have a dataset with multiple class attributes, and you want to
process it with Weka, it needs to be split into several datasets, one
for each class attribute.

Eibe

On Jul 18, 2005, at 7:21 AM, Surendra Singhi wrote:

> Michael Thon <[hidden email]> writes:
>
>> Greetings - I have a multiclass dataset that I imported into weka.  I
>
> If I understand you correctly, then you mean dataset with instances
> having
> more than one class.
>
>> would like to try the multiclasslassifier with naiveBayesSimple .  It
>> seems to be returning only 1 classifier (only one set of
>> probabilities)
>> for the class selected in the dropdown list on the left side of the
>> screen.  How can I select multiple classes to be trained?
>
> The term multiclassclassifier as used in Weka is misleading, if you
> read the
> documentation; it says
> "Class for handling multi-class datasets with 2-class distribution
> classifiers"
>
> what it means, is that this(MCC) classifier can be used to handle
> datasets
> with more than 2 classes by using binary classifiers(classifiers which
> take
> only 2 classes). For example say a classification problem with 10
> classes, and
> you want to make use of SVM which is a binary classifier.
>
> So, getting back to your problem, currently it is not possible to do
> multi-class classification using Weka, unless you extend Weka.
>
> Eibe, please correct me, if I am wrong.
>
> Cheers.
> --
> Surendra Singhi
> http://www.public.asu.edu/~sksinghi/index.htm
>
> Great wits are sure to madness near allied,
> And thin partitions do their bounds divide.
>
>    (John Dryden, Absalom and Achitophel, 1681)
>
>
> _______________________________________________
> Wekalist mailing list
> [hidden email]
> https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist


_______________________________________________
Wekalist mailing list
[hidden email]
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist