I have a general question. I have a very non-stationary time series data. The DNN gives a very bad result (well, I did not try a LSTM yet). My question is, how the defferencing
