[R] RWeka and Back Propagation NN

Patrick Connolly p_connolly at slingshot.co.nz
Tue Mar 26 08:43:07 CET 2013


On Sat, 23-Mar-2013 at 10:10PM +0100, Rui Esteves wrote:

|> Hello,
|> 
|> I have a trained Back Propagation Neural Network model in weka.  I
|> would like to re-evaluate the NN using R with a given input.  How
|> can I do this? I could not find an example of RWeca that applies to
|> NN

As the documentation states, there's not a lot of output from Weka
that can be used in R.  If Weka produces an arff file, that's easily
read into an R dataframe which is useful for doing decent plots.  If
you're good at writing perl scripts you might be able to do some neat
tricks with Weka's verbose output.

Often I've found that the only way to get Weka output from R is to use
system() calls to run native Weka commands.  Some options can be
passed in list form and when that works it's quite elegant, but even
so sometimes I've had to resort to system() calls even though similar
cases do work.  Where RWeka is very useful is producing those big long
headers in arff files from R dataframes and it's worth having just for
that.  And as I stated above plotting Weka results.

I've not had experience with Weka's neural networks so I don't have
much of an idea of what you have and what you want to do with it, but
the above might give you some ideas.

HTH

-- 
~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.   
   ___    Patrick Connolly   
 {~._.~}                   Great minds discuss ideas    
 _( Y )_  	         Average minds discuss events 
(:_~*~_:)                  Small minds discuss people  
 (_)-(_)  	                      ..... Eleanor Roosevelt
	  
~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.



More information about the R-help mailing list