mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Loic Descotte <loic.desco...@kelkoo.com>
Subject SGD 20 newsgroup classification full example
Date Tue, 13 Sep 2011 10:13:33 GMT
Hi,

I'm currently studying the 20 newsgroup example for classification in 
Mahout in Action book.

In the naive bayes example, train et test are both explained.

In the SGD example (chapter 14), only train is explained. The Welford 
algorithm is shown to evaluate the learning model, but is not used on 
train data : the files from "20news-bydate-test" folder dataset is never 
used (it's only used with naive bayes example), only  
"20news-bydate-train" is called in the Java program.

I try to adapt the naive bayes test code (from 
org.apache.mahout.classifier.bayes.TestClassifier) but it seems very 
specific to Bayes Algorithm.

Do you kwnow where I can find a full SGD example for this kind of problem?

Thanks

Loic

Ps : I also replied to this topic on manning forum, sorry if it 
duplicates the question :
http://www.manning-sandbox.com/thread.jspa?messageID=118179&#118179

Mime
View raw message