mahout-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Karl Wettin <>
Subject cross fold validation
Date Thu, 14 Aug 2008 22:34:12 GMT
What's the use of distributed cross fold validation if the  
classification algorithm already is scalable? I suppose it can make  
sense if there are more nodes available than the classification  
algorithm can use. Is the number of maximum nodes a computable value?

Alternative question: in what simple ways should a cross fold  
validation be configurable in order to work as great as possible with  
any given calssification algorithm?


View raw message