How to generate predictions with NARX?

Hello,

I have a very simple question regarding NAR usage. I understand from the tutorials how to train a neural network on given target series. Can somebody confirm that my approach to predict future values in out of sample test set is correct?

targetSeries = tonndata(C,true,false); feedbackDelays = 1:5; hiddenLayerSize = 10; net = narnet(feedbackDelays,hiddenLayerSize); [inputs,inputStates,layerStates,targets] = preparets(net,{},{},targetSeries); net.divideParam.trainRatio = 70/100; net.divideParam.valRatio = 15/100; net.divideParam.testRatio = 15/100; [net,tr] = train(net,inputs,targets,inputStates,layerStates);

C2=[a,b,c,d,e] (a,b,c,d,e are five consecutive values from out of sample test set) targetSeries = tonndata(C2,true,false); [xs,xis,ais,ts] = preparets(nets,{},{},targetSeries); ys = nets(xs,xis,ais);

Will ys be the prediction generated for the sixth value that comes after a,b,c,d, and e?

Thank you very much.

[Greg Heath]

Why don't you just try the code on some of the MATLAB timeseries example data ???

 help nndatasets

 doc nndatasets

In particular, the simplenar_dataset!

If you have problems you can post details and we can help.

Hope this helps.

Greg