Interpolation using Neural Networks

Hey guys,

so my problem is as follows:

I am trying to interpolate the hourly data points between daily measurement data of the wind speed.

I do have the hourly wind speed from 46000 measurement points for one year to use for training,validation etc.

So essentially what I am trying to do is get 24 data points from my neural network and together they should have the same mean as the "real" data.

Since I am beginner concercong neural networks I am not sure how to begin.

Would you say that it is a fitting problem or more of a time series problem?

I am thankful for your help and for all suggestions!

Right now I am using the neural network Matlab Toolbox.

Answer 1

[Greg Heath]

Plot the data.

Look at the plots

Are there regions that are seasonal?

Separate the data that is relevant to your problem

Can you subsample the data and preserve the shape?

Obtain a timeseries model and predict

Hope this helps.

THANK YOU FOR FORMALLY ACCEPTING MY ANSWER !

Greg

[Lukas Wammes]

Hey Greg,

Thanks a lot for your help.

I can't subsample the data actually.

What keeps me up is how I can make sure that the predicted data has the same mean(per day in this case) as my actual known data.