Find optimal values ​​of the network error and measurement NarX

Could someone resolverme any of these questions?

I've been testing my code in dao indications pollution_dataset with me and I have several questions that I would like:

One-variable at different scales. Normalize Where pollution_dataset normalized variables?

3 NARXNET entries are not optimal. Important discovery AUTO AND CROSSCORRELATIONS. H vary if necessary and do this step?, Trains and observed network autocorrelation graphs and cross correlation, and notes that the errors are within the confidence limits (dotted red line)?

4 DO NOT USE THE DEFAULT DIVIDERAND dividetrain I tried also but the data does not allow me to divide by train, validation, test because I guess in the FAQ, then train with all lenses. Still, I think that has a lot of errors. I'd like to tell me how to use dividetrain.

5 INITIALIZE RNG running the OS can double as I do and when? After training and simulation openloop?. I use RNG (sprev)?

6 Normalize MSE VARIANCE WITH THE MIDDLE AND GET META determining the coefficient R ^ 2 I read the wikipedia and not sure how apartado.Habría some code where I could see? I find it hard to understand.

7-H ~ Hub/10 not understand this indication.

If you know of a place where he could see the complete code on these questions I'd appreciate it, pollution_dataset applied to other data in order to understand better. I sent an email with my questions more Clearly.