How to make an out-of-sample forecast from a MLPRegressor with Time Series Data?

Question

I would like to make out-of-sample predictions with a neural network model, but I don\'t know how to make an out-of-sample forecast.

What I did was explore a Pipeline