Automatic demand forecast model selection in supply chains: a forecast value-added analysis of selection strategies, machine learning, and hyperparameter optimisation

Key Points

• Forecast value-added analysis shows enhanced accuracy in demand forecasting, proving selection strategies vital.
• Key evidence indicates that optimized machine learning models can significantly reduce forecast error rates.
• Assessment across various selection strategies reveals effectiveness of hyperparameter optimisation in model tuning.
• Highlights the necessity for ongoing development in forecasting systems, emphasizing the role of innovative strategies.