<![CDATA[Monthly palm oil production fluctuates due to seasonal patterns and long-term trends, making accurate forecasting essential for operational planning in plantation companies. This study aims to compare seven Holt–Winters Exponential Smoothing models with different trend and seasonal component configurations to forecast the monthly palm oil production of PT XYZ using a univariate approach. The analysis is based on 135 observations covering the period from January 2015 to March 2026. The performance of each model configuration was evaluated using the last 12 months as the test set and assessed based on the Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE). The results indicate that the Holt–Winters model with a multiplicative trend and additive seasonality consistently achieved the best forecasting performance, with a MAPE of 5.71%, an MAE of 284.89 tons, and an RMSE of 331.83 tons, substantially outperforming the other six model configurations. The selected model was subsequently used to generate production forecasts for the next 12 months, providing a basis for managerial decision-making in harvest planning, mill capacity management, workforce allocation, and sales strategy.]]>
Copyrights © 2026
