how to get the best the number of hidden neurons and layers for a NARXNET?

i want to know if there is a general way of choosing the correct or right number of hidden neurons and delays in a NARX to obtain the best training

Answer 1

[Greg Heath]

There are zillions of my posts in both the NEWSGROUP and ANSWERS that use the criterion

       Ntrneq >= Nw      %( number of training equations >= number of unknown weights )

to obtain an upper bound, Hub, on the number of hidden nodes that should be used without early stopping and/or regularization .

Search both using

 greg narx Hub

or

 greg narxnet Hub

Hope this helps.

Thank you for formally accepting my answer

Greg