Predict futures values in ntstool (NAR)

I have used ntstool to create a neural network say net . I am using NAR. I have a vector v of size 271 x 1 having data of an element at 271 timesteps. I am assuming d to be 5 , i.e, value at each step is dependent of 5 previous steps. Now I need the value of my element at 272th step. What is the syntax to do do?

A peculiar observation is the command sim(net,t); or net(t); where t is any row vector returns me same value independent on the choice of vector t.

Here is my training function:

    function results = time_series(input,hiddenLayerSize,train_function, validation_no,test_no,time_step,iterations)
   targetSeries = tonndata(input,false,false);

    feedbackDelays = 1:time_step;
    net = narnet(feedbackDelays,hiddenLayerSize);

    % Choose Feedback Pre/Post-Processing Functions
    % Settings for feedback input are automatically applied to feedback output
    % For a list of all processing functions type: help nnprocess
    net.inputs{1}.processFcns = {'removeconstantrows','mapminmax'};

   [inputs,inputStates,layerStates,targets] = preparets(net,{},{},targetSeries);

    net.divideFcn = 'dividerand';  % Divide data randomly
    net.divideMode = 'time';  % Divide up every value
    net.divideParam.trainRatio = (100-validation_no-test_no)/100;
    net.divideParam.valRatio = validation_no/100;
    net.divideParam.testRatio = test_no/100;

   net.trainFcn = train_function;  % Levenberg-Marquardt

   for i= 1:iterations
    [net,tr] = train(net,inputs,targets,inputStates,layerStates);
    end;

    % View the Network
    %view(net);

    if(closed_loop)
    netc = closeloop(net);
    [xc,xic,aic,tc] = preparets(netc,{},{},targetSeries);
    yc = netc(xc,xic,aic);
    perfc = perform(net,tc,yc);
  nets = removedelay(net);
    [xs,xis,ais,ts] = preparets(nets,{},{},targetSeries);
    ys = nets(xs,xis,ais);
    closedLoopPerformance = perform(net,tc,yc);
    end

    results.net = net;
    results.inputs = inputs;
    end

Answer 1

[Greg Heath]

1. The best way to get help is to run your code on the command line with one of the MATLAB example data sets. See

    help narnet
    doc narnet

    or

    help nndatasets
    doc nndatasets

Then post questions with supporting relevant code, error messages and results.

2. Find the delays from the significant delays of the autocorrelation function

3. Find the number of hidden nodes by trial and error. Do not use more than you need.

4. Use divideblock, NOT dividerand to keep from ruining the time correlation structure.

5. Search for posted examples in the NEWSGROUP, then in ANSWERS. Use the searchwords

   greg narnet

Hope this helps.

Thank you for formally accepting my answer

Greg