I believe a simple moving average would achieve the same result and that mixing training and test data for validation invalidates your model results... On a po

I believe a simple moving average would achieve the same result and that mixing training and test data for validation invalidates your model results... On a positive note, nice showcase of how simple implementation with Keras is.