Re: MISMO Classifier: how to set Instance Weights?

Am 07.01.24 um 20:49 schrieb Peter Reutemann:
> 
> My proposed workaround disables capabilities checks *globally*, which
> can have unwanted side-effects when providing incorrect data. Hence to
> be used with caution.

Well, what can I say, it worked!

However, why does the classifier use the PropositionalToMultiInstance 
filter at all? My ARFF file is already in the correct multi instance 
format with Bags etc. as specified in the docs.

> Just my 2c from way back when I was working on propositionalization...
> - The MIWrapper usually outperformed all other MI methods.
> - Applying the RELAGGS to your data could also be an option to explore:
>    https://weka.sourceforge.io/doc.stable/weka/filters/unsupervised/attribute/RELAGGS.html

I have also gotten decent results from the MIWrapper, which (I would 
believe) needs to use the PropositionalToMultiInstance filter internally 
as well, but for some reason it does not run into the problem that the 
MISMO classifier experiences.

I am probably missing some fundamental knowledge here. My code is not 
employing any filters at all, just using the classifiers. It is hard to 
understand which filters the classifiers use internally and how to, say, 
configure them to use RELAGGS instead (if that would lead to higher 
accuracy).

Many thanks for your help,
Ulrich

_______________________________________________
Wekalist mailing list -- [email protected]
Send posts to [email protected]
To unsubscribe send an email to [email protected]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html