You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@mahout.apache.org by Adam Baron <ad...@gmail.com> on 2012/10/12 20:54:02 UTC

Is there something like clusterdump for Naïve Bayes?

I'm somewhat new to Mahout and am running through the Mahout 0.7 examples.
 So far I've gone through classify-20newsgroups.sh for Naïve Bayes
and cluster-reuters.sh for K-Means Clustering.  I really like
the clusterdump function in the K-means example because it helps me
understand what the main drivers behind each cluster are.  Is there a
similar function anywhere to see the top drivers behind the Naïve Bayes
classification in the naiveBayesModel.bin file?

I suppose I could write some of my own code to parse out the alphaI,
weightsPerFeature, weightsPerLabel, perLabelThetaNormalizer values from
the naiveBayesModel.bin similar to code in
the org.apache.mahout.classifier.naivebayes.NaiveBayesModel class.
 However, I'd much rather reuse someone else's code if it exists already.

Thanks,
          Adam