MATLAB: How to time series forecasting with the trained neural network

Deep Learning Toolboxneural networksimsimulationtime series forecasting

Hello ! I would like to train neural network and then on that neural network do time series forecasting. In orderd to do so I use such rather typical code.
[x,xi,ai,t] = preparets(net,{},{},T);
net.divideFcn = 'dividerand';  % Divide data randomly
net.divideMode = 'time';  % Divide up every value
net.divideParam.trainRatio = 70/100;
net.divideParam.valRatio = 15/100;
net.divideParam.testRatio = 15/100;
[net,tr] = train(net,x,t,xi,ai,'useParallel','yes');
y = net(x,xi,ai);
e = gsubtract(t,y);
mse = perform(net,t,y);
when I have trained naural net I would like to get new y(t+1) value
output = net(y(t)) //or any number
output = sim(net,[y(t-1) y(t-2)]) //or vector of any numbers
Unfortunately doesn't matter what is the value of y(t) I get always output = -1.8812.
Could anybody explain me what I do wrong? What i should do to be able to predict y(t+1) value.
Thank you very much for helping. Janek

Best Answer

 [ Y Xf Af ]= net( X, Xi, Ai );  % In the target region
for post target prediction
 Xpi = Xf
 Api = Af
 [ Yp Xpf Apf ]= net( Xp, Xpi, Api );
Hope this helps
Thank you for formally accepting my answer
Greg
Related Question