Hi Nagaraj, the partial data implementation of Decision Forest is still a "work in progress", the next code update (wich I hope will come next week) should improve the accuracy of the generated classifier and fix a bug related to categorical attributes. And yes, I'll add some documentation about using the partialdata imp.
--- En date de : Mar 29.12.09, Nagaraj K <[email protected]> a écrit : > De: Nagaraj K <[email protected]> > Objet: PartialData Random Forest usage. > À: [email protected] > Date: Mardi 29 Décembre 2009, 17h46 > Hi, > > I am trying to use the partialdata mapreduce Random Forests > feature > from the latest svn branch. > https://issues.apache.org/jira/browse/MAHOUT-145 > Our data has ~ 100-200M points with 8-10 categorical > attributes. I > tried looking for a sample usage for this map-reduce > version, but > could not get all the parameters. > > Could you please let me know, if there is any documentation > or code > available on this. > > -- > Thanks, > Nagaraj K >
