narx - inputs and targets

In the NARX algorithm (Neural Net Toolbox), could someone explain what the difference between inputs and targets. I've read the manual, help guide, and have looked at the two arrays but still do not understand. Why would we include targets if this is what the algorithm is trying to predict?

Also is there a way to run a test case to via NARX and check prediction capability (R-squared...)?

Thank you so much for your help.

Answer 1

[Greg Heath]

You cannot build a successful net without knowing what transformation you want it to perform.

In general, inputs are typical examples from a known distribution or collection of inputs.

Targets are the known result of the transformation applied to specific inputs.

The purpose of the trained net is to be able to GENERALIZE, i.e., perform the transformation on arbitrary examples from the distribution or collection of examples.

I have posted zillions of examples using the collection of MATLAB examples obtained by using the commands

 help nndatasets
 doc nndatasets

 Search BOTH the NEWSGROUP and ANSwERS using

 greg narxnet tutorial

and

 greg narxnet

Hope this helps

Thank you for formally accepting my answer

Greg