earticle

영어

Data-driven decision-making is essential in the Fourth Industrial Revolution. In particular, accurate demand forecasting plays a critical role in optimizing supply chains and improving responsiveness in the global trade environment. This study aims to compare the forecasting performance of two models: ARIMA (Auto-Regressive Integrated Moving Average), a traditional time series method, and GRU (Gated Recurrent Unit), a deep learning-based approach. Using real-world demand data from the automobile parts industry, both models were evaluated based on MAPE (Mean Absolute Percentage Error), RMSE (Root Mean Squared Error), and RMSLE (Root Mean Squared Log Error). The results show that the GRU model consistently outperforms the ARIMA model, especially for highly volatile and irregular (lumpy) demand data. These findings suggest the practical applicability of GRU in demand forecasting for industries dealing with irregular patterns, and highlight the potential of combining traditional and deep learning methods for improved forecasting accuracy in international trade settings.