A robust and resilience machine learning for forecasting agri-food production

Area of Science:

• Agricultural Economics
• Operations Research
• Supply Chain Management

Background:

• Agri-food production systems face increasing disruptions and risks.
• Existing capacity planning models often lack comprehensive consideration of resilience and robustness.
• Effective risk management is crucial for stable agri-food supply chains.

Purpose of the Study:

• To develop and present a novel framework for agri-food capacity production that integrates resiliency and robustness.
• To address disruption and risk factors in agri-food production planning for the first time.
• To provide agri-food managers with a mathematical method for improved forecasting and production management.

Main Methods:

• Application of robust stochastic optimization techniques.
• Incorporation of robustness into the constraint's objective function and the resiliency situation.
• Minimization of mean absolute deviation and coefficient of standard deviation errors using a linear function.
• Comparison of the proposed model (Robust and Resiliency Mean Absolute Deviation - RRMAD) with sine-type functions.

Main Results:

• The proposed RRMAD model achieved a 1.28% lower value compared to other sine-type functions.
• Analysis of the impact of varying conservativity coefficient, confidence level, weight factor, resiliency coefficient, and scenario probability.
• Demonstrated trade-offs between RRMAD, R-squared, weight factor, and resiliency coefficient.

Conclusions:

• The developed mathematical framework offers a superior approach to agri-food capacity production planning.
• The RRMAD model enhances decision-making by providing more accurate forecasts and risk-aware production strategies.
• Agri-food managers can leverage this method to optimize production and mitigate potential disruptions.