Traditional methodologies for time series prediction take the series to be predicted and split it into Q1 training, validation, and test sets. The first one serves to construct forecasting models, the second set for model ...