Machine Learning Demand Forecasting Based on ARIMA and Support Vector Machines to Improve Supply Chain Efficiency

This study proposes a hybrid forecasting method that combines ARIMA and SVM models to address the complexity of supply chain demand forecasting. The ARIMA model is used to capture linear trends and seasonal fluctuations in time series, while the SVM model is responsible for modeling the nonlinear residual part. By combining the advantages of both, this paper builds a more accurate demand forecast model. Experimental results show that the combined model performs well when dealing with large market fluctuations and non-linear demand changes. Compared with the ARIMA and SVM models applied individually, the forecast error is significantly reduced. The model shows good prediction accuracy in multiple categories such as food, household products and electronic products, provides a more comprehensive demand forecast solution for supply chain management, and helps improve the company's inventory management and market response capabilities.