Perfomance value of a neural network.

I solved my time series problem using NARX network. I am getting very low performance. how can i improve it???

 What are the values of MSE that are good for a network. how should the graphs of correlation should look.

any source how to read these graphs???

Answer 1

[Greg Heath]

If

 MSE00 = mean(var(t',1))& Average target variance

a useful training goal is

 NMSE = mse(t-y)/ mean(var(t',1)) <= 0.01

therefore

net.trainParam.goal = 0.01*MSE00

i.e. the net models more that 99% of the average target variance.

Search NEWSGROUP and ANSWERS using

 greg MSEgoal

Hope this helps

Thank you for formally accepting my answer

Greg