Evaluating a Keras prediction

DanelMadariaga

Hello,

I have set up and trained a Keras model predicting some values of a dataset (a time series of slowly evolving continuous variables).

When I evaluate it, I get a modified dataset with the prediction of the target variable added on the rightmost column (and optionally, deciles and absolute and relative errors for the prediction).

But, instead of just getting a prediction for the mean of that value as a function of time, how could I get also a prediction for its standard deviation, so that I can have a confidence interval for estimating the goodness of the prediction?

Thanks in advance for your help, kind regards,

Danel Madariaga