Prediction Model of Empty Fruit Bunch Production at PKS Sawit Hulu, Langkat Regency Based on Historical DataUsing the Arima Method

Mayah Sapriani (Universitas Malikussaleh)
Muchlis Abdul Muthalib (Universitas Malikussaleh)

International Journal of Engineering, Science and Information Technology
Vol 6, No 2 (2026)

This study aims to develop a forecasting model for Empty Fruit Bunch (EFB) production at PKS Sawit Hulu, Langkat Regency, using the Autoregressive Integrated Moving Average (ARIMA) method based on historical production data. The dataset consists of daily time series data collected from 2021 to 2025. The research process includes data collection, preprocessing, data cleaning, stationarity testing, parameter identification using the Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF), ARIMA model development, forecasting, and model performance evaluation using Mean Absolute Percentage Error (MAPE) and Mean Squared Error (MSE). The evaluation results show an MSE value of 4,807,984,891.58 and a MAPE value of 29.32%. Based on model comparison and residual diagnostic testing, ARIMA (1,0,1) was selected as the best model because it produced lower error values and satisfied the white noise assumption. These results indicate that the model is adequate for representing the pattern of EFB production data and can be used for short-term forecasting. The forecasting results also show that EFB production fluctuates over time in line with changes in the amount of Fresh Fruit Bunches (FFB) processed during each period. These fluctuations are influenced by operational conditions, processing activities, and the availability of raw materials. The developed model is expected to assist PKS Sawit Hulu in production planning, waste management, storage capacity planning, biomass utilization, and operational decision-making. Overall, the ARIMA model provides a practical, systematic, and reliable approach for forecasting industrial production data based on historical patterns accurately.