Determine impact of Neural Network inputs on outputs.

I am developing a neural network to forecast water consumption. I am using weather parameters but I only want to use input parameters that correlate well with the output. For example, I assume that rainfall would correlate well but dew point may be better. How can I determine which input is best?

If I use the phase of the moon as an input, how can I prove that this input contributes nothing and may actually insert noise to the network.

Is it possible to determine the impact of individual input parameters on the output(s) of the Neural Network using the Neural Network Toolbox or do I need to use some other toolbox?

Answer 1

[Greg Heath]

See the folowing post:

View weights used in a neural network created after using the GUI

Sarah asked on 18 Jun 2012 at 10:11

Latest activity: Answer by Greg Heath on 19 Jun 2012 at 1:43

[Walter Roberson]

http://www.mathworks.com/matlabcentral/answers/41421-view-weights-used-in-a-neural-network-created-after-using-the-gui