XGBoost with Q-learning for complex data processing in business logistics management

Abstract

The modern business landscape is characterized by complex technical information, economic globalization, and high customer expectations. These factors have led to significant changes in various industries. To ensure customer satisfaction, companies rely on supply chain management (SCM) for the timely delivery of products and gathering feedback for analysis. The collected customer data is often complex and requires advanced methods for processing and management. To effectively manage demand and supply in real-time, businesses must have the ability to handle complex data. Due to the inefficiency and ineffectiveness of traditional methods for the increased data volume and speed, much-emerging research is being conducted on how to harness complex data in SCM. This paper examines the limitations of conventional methods and introduces an Artificial Intelligence (AI) approach based on Q-Learning algorithm with Extreme Gradient Boosting (QL-XGB) model. The QL-XGB method is applied to select suppliers and predict their future demand for the production of products. It is built on the foundation of accurate data and analysis of supply chain characteristics using metrics such as MAE and RMSE. The results show that the QL-XGB model with an accuracy rate of 96.02% outperforms QL and XGB models with respective accuracy rates of 93.44% and 94.68%.