[Corpora-List] Decision tree : maximise recall over precision

Emmanuel Prochasson emmanuel.prochasson
Wed Apr 22 15:08:04 CEST 2009


Eric Atwell a écrit :
> Enmmanuel,
>
> Surely a good decision procedure is "JUST SAY NO!" - "only" 99.9%
> accurate! I wish PoS-taggers and other text annotation tools were as
> good!
>
> It sounds like you want to find out how to set a WEKA decision-tree
> builder to NOT prune any branches ... this question is better put to
> the WEKA mailing list wekalist at list.scms.waikato.ac.nz - see
> https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist to join

Thank you very much for all your answer, I'm glad to see this matter is of interest and that it does not seem to be straightforward (though, it would have been convenient).

This is not exactly a weka-oriented question (it would have been if there were a straightforward solution) though I'll seek for an answer, or at least some clue on the mailing list.

I'll dig the question with all I gathered. Maybe it's very simple using brute force or hacked algorithm, maybe it can be done with a formal, theoretical-based modification to Decision Tree Learning algorithm. I'll keep the list in touch of what I get.

Thank you very much,

-- Emmanuel Prochasson



More information about the Corpora mailing list