Research on Supply Chain Inventory Demand Forecasting Model Based on LightGBM

Inventory demand forecasting is a core component of supply chain management and serves as a critical foundation for inventory control, cost reduction, and operational coordination. Accurate demand prediction enables enterprises to balance supply and demand more effectively, mitigate the risks of overstocking or stockouts, and enhance overall decision-making efficiency. With the increasing complexity and volatility of market demand, traditional forecasting methods often struggle to capture nonlinear patterns and dynamic fluctuations in real-world data. To address these challenges, this paper proposes an inventory demand forecasting approach based on the Light Gradient Boosting Machine (LightGBM) model. Using retail sales data obtained from a publicly available dataset, the proposed method constructs a demand prediction model that accounts for complex temporal characteristics such as seasonality and trend variations. The forecasting performance of LightGBM is systematically compared with that of autoregressive integrated moving average models, gradient boosting-based models, and linear regression approaches. Experimental results demonstrate that the LightGBM-based method achieves superior forecasting accuracy while maintaining higher computational efficiency, particularly when dealing with data exhibiting strong nonlinearity and mixed temporal patterns. The findings indicate that the proposed approach is both effective and robust, offering a practical solution for inventory demand forecasting in supply chain management and providing meaningful support for data-driven inventory optimization and operational planning.