nnet

[Carlos Crosetti]

Ho, I am playing with Rattle with a dataset of 8 input variables and one response variable. All input variables are numeric.

In the model tab, I see a control where I dan set the nodes of a hiden layer, when the evaluation report is shown I see this label:

A 8-5-1 network with 59 weights.

How this is niterpreted?

I assume 8 is the number of neurons in he input lyer, 1 is the output layer single neuron.

Is 5 the number of nodes in a single hidden layer  or 5 is the number of hidden layers?

What is exactly the meaning of

eights for node h4:

 b->h4 i1->h4 i2->h4 i3->h4 i4->h4 i5->h4 i6->h4 i7->h4 i8->h4 

  0.49   0.56   0.44   0.41   0.51   0.38   0.22   0.47  -0.41 

Is there a R command that could draw the topology of the current network in use?

Thanks, Carlos

[Graham Williams]

The Rattle window includes a legend to explain the symbols.

There may be a package to display the topology - if you find one let me know and I will add it to Rattle!

Thanks.

Regards,
Graham

--
You received this message because you are subscribed to the Google Groups "rattle-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rattle-users+unsubscribe@googlegroups.com.
To post to this group, send email to rattle...@googlegroups.com.
Visit this group at https://groups.google.com/group/rattle-users.
For more options, visit https://groups.google.com/d/optout.

[Carlos Crosetti]

Thanks Graham.

I think the neural net plot will be very useful and will set the neural net at the same graphic display level other models exhibit.