Hi - I would like some advice on how data are treated in the Neural Network module. My response data are species richness, they range from 0 to 78 (n=9000). I have a number of X variables (up to 38) that I am trying to use in an MLP regression NN.

1. the STATISTICA instruction video https://www.youtube.com/watch?v=7phbGIqPOLY&list=PLB804A810436AFB03&index=26 at 1:08 minutes it says that input data are normalized 0 - 1.

2. So if input data (response data - Y) are normalized to 0 - 1, then how do the activation functions such as **Identity** (with an output range of -infinity to +infinity) or **Negative** **Exponential** (with an output range of 0 to +infinity) apply these normalized response data with a range of 0 - 1.

3. Given my response data range from 0 to 78, it seems to me that they would only fit the **Negative** **Exponential** function in both the hidden layer and the output layer. But if they have been re-scaled from 0 to 1, then should I be using **Logistic** or **Sine** (both range from 0 to 1)?

I just don't understand which activation function applies to response data that range from 0 - 78?

4. Should I be normalizing my predictor (X) inputs to 0-1 as well?

Regards