Integrating Holt-Winters Exponential Smoothing and SAW for Distribution Priority Optimization

Abstract

Effective distribution planning requires both accurate demand forecasting and objective prioritization mechanisms to ensure operational efficiency and service reliability. At CV. Citra Nalar Teknologi, distribution scheduling was previously conducted manually without incorporating predictive demand analysis, resulting in inconsistent and subjective prioritization decisions. This study proposes an integrated decision support framework that combines Holt–Winters Exponential Smoothing (HWES) for demand forecasting and Simple Additive Weighting (SAW) for multi-criteria distribution ranking. Monthly sales data from January 2024 to December 2025 (24 periods) were analyzed using the additive HWES model to capture level, trend, and seasonal components. The forecasting results achieved MAPE values ranging from 3.82% to 7.47%, supported by low RMSE, MAE, and sMAPE, indicating high predictive accuracy. Comparative evaluation further shows that HWES outperforms baseline methods, including Moving Average and Seasonal Naive. The forecasted demand was then incorporated as a primary criterion in the SAW model, alongside delivery distance, customer priority level, and stock availability. The proposed framework was evaluated using two sets of actual customer orders (January 4–5 and January 6–7, 2026). The results demonstrate that the integration produces consistent, data-driven distribution priorities and improves decision objectivity and transparency compared to manual scheduling. This study contributes by integrating time series forecasting with multi-criteria decision-making into a unified framework for practical distribution optimization.