The Optimization and Contracting of Supply Chain with Machine Learning-Based Demand Forecasting

This study addresses the optimization of contract decisions in supply chain management by comparing classical and machine learning-based demand forecasting methods. The classical method assumes that demand follows a normal distribution, with the mean and standard deviation calculated for each product type from historical data. We then employed three machine learning algorithms-linear regression, random forests, and XGBoost-to predict demand more accurately and evaluate their effect on supply chain profitability. The main objective of this research is to evaluate whether machine learning models can outperform traditional forecasting methods in terms of supply chain optimization. The results demonstrate that machine learning methods, especially XGBoost, significantly improve forecasting accuracy and lead to higher profitability compared to the classical normal distribution approach. Among the machine learning models, XGBoost provided the most accurate predictions, as measured by performance metrics such as Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R-squared (R۲). The study finds that using machine learning-based demand forecasting results in better decision-making for inventory and pricing strategies. In conclusion, this study shows that incorporating machine learning techniques, particularly XGBoost, into contract optimization models can enhance decision-making processes, leading to more efficient supply chain management. The findings suggest that these techniques offer substantial advantages over classical methods, enabling businesses to achieve higher profitability and responsiveness. Further research could explore the integration of additional features and other machine learning models to improve forecasting accuracy in complex supply chain environments.

کلیدواژه ها:

Supply Chain Coordination ، Machine Learning ، Inventory Optimization ، Contracts ، Artificial Intelligence