neural time series tool prediction

Hi,

I have kind of a trivial question. I just started working with neural networks and have really enjoyed using the NN toolbox. Recently I've been dealing with some data that definitely calls for the time series tool. I train up the network with n days of data and I get beck the trained network, along with some performance information. My question is, how do I take that network and get a prediction for n+1th (I'm actually interested in the n+30th, but same question) step? I'm totally puzzled by the use of the 'preparets' function.

Thanks,

Jordan

[Greg Heath]

What does "I train up the network mean"? To train a time-series prediction net you have to give it input/target pairs. So, for a step ahead prediction, the input is p(:,n-d:n) and the corresponding output is t(:,n) = p(:,n+1). A decent value of d can usually be estimated by looking at the autocorrelation function.

Sorry, I can't help you with the rest because my computer

with the latest version of MATLAB isn't working.

Greg