Enhancing Supply Chain Resilience through Machine Learning-Based Predictive Analytics for Demand Forecasting

Abstract

In the era of global supply chain complexities, ensuring resilience in supply chain operations is critical for minimizing disruptions and maintaining efficiency. Traditional demand forecasting methods are unable to adapt to changing market conditions, which leads to inefficient inventory control and resource allocation. In the retail industry, precise demand forecasting is crucial for maximizing inventory control, reducing stockouts, and enhancing financial decision-making. This study employs Extreme Gradient Boosting (XGBoost) to enhance sales prediction accuracy using Walmart sales data. The dataset is partitioned into training and testing sets in an 80:20 ratio, and the XGBoost model is fine-tuned to achieve optimal performance. Experimental results indicate superior predictive accuracy, with an R² of 95.51%, a minimal MAE of 0.0024, and an MSE of 4.79. Visualization techniques, including density curves and correlation heatmaps, provide deeper insights into feature relationships and data distribution. The findings demonstrate the robustness of XGBoost for demand forecasting, offering a data-driven approach for retailers to enhance operational efficiency and strategic planning.