Fig. 1
From: Primacy of feature engineering over architectural complexity for intermittent demand forecasting
End-to-end experimental workflow for intermittent demand forecasting, illustrating data preprocessing, SHOS feature generation, model training, and rolling-window evaluation.
