Forecasting of Palm Oil CPO Production Results at PTPN III Batang Toru Plantation Using The Autoregressive Integrated Moving Average Method

Abstract

The increasing demand for palm oil as a raw material for food and energy industries has driven the need for accurate forecasting methods to optimize palm oil production management. This study aims to forecast Crude Palm Oil (CPO) production at PTPN III Batang Toru Plantation using the Autoregressive Integrated Moving Average (ARIMA) method. Monthly time series data from January 2020 to January 2024, including Fresh Fruit Bunches (FFB), loose fruit, and CPO yields, were analyzed to build the forecasting model. The Augmented Dickey-Fuller (ADF) test confirmed that the data is stationary without differencing. Based on the ACF, PACF, and white noise tests, the ARIMA(1,0,1) model was identified as the best fit. The forecasting results indicated a potential increase in CPO production from January 2025 to December 2026. However, alternative models like CPOF showed poor accuracy, with a high MAPE of 442.12%, suggesting the need for further model refinement. Despite limitations, the ARIMA method remains effective for short-term forecasting and supports data-driven decision-making in the plantation sector.