Can I use the network from 'nftool' output to predict other data?

Hello,

This is first time I am trying neural network. Following the tutorials (videos) in the mathworks website, I used the 'nftool' to create a relationship between a set of input parameters (e.g., predictors or x1, x2, x3, ...xn in regression) and a target parameter (e.g., predicted or y in the regression). Now, can I use the network (or relationship) created here to predict the value (e.g., y) using another independent set of same predictors? For example, I have a file with only predictors (x1, x2, x3, ...xn). Can I use these to predict y (as we develop a regression equation and use the relationship to predict the y value from another set of data). How can I express the relationship and where I can get that in the output? Can I express the relationship as we express in the regression equation? Thank you so much. I really appreciate your help.

I am using matlab version 7.9.0 (R2009b) and the neural netwrok toolbox version 6.0.3.

Thanks a lot in advance.

Asim

Answer 1

[Greg Heath]

If the old and new data can be considered to be samples from the same probability distribution, then

ynew = net(xnew);

should do the trick.

Answer 2

[Dim Lit]

I have somehow the same question with the aforementioned one! Lets assume that i want to use the neural network tool in order to correlate my data (a whole set of input and the target data set as well) so as to produce an analytical solution of all these data to predict sth computationally without to use again and again the matlab tools. My question is: Is there any way to take the basic equation which the matlab tools use (after all the training processes etc.) to produce the "analytical" solution?? Thnx in advance DL