SARIMA MODELING FOR RAILWAY FREIGHT TRANSPORTATION FORECASTING ON SUMATRA

Putri Fauziahtul Asri (Universitas Negeri Padang)
Mutia Yollanda (Universitas Andalas)
Windry Novalia Jufri (Universitas Negeri Padang)
Jonni Mardizal (Universitas Negeri Padang)

Mathematics and Applications (MAp) Journal
Vol 8, No 1 (2026)

This study develops a time series model to forecast railway freight volume in Sumatra using monthly data from January 2013 to June 2025. A seasonal autoregressive integrated moving average (SARIMA) model with a drift component captures both trend and seasonal patterns in the data. Model selection is based on the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC). The results show that the SARIMA(0,1,2)(1,0,0)[12] with drift provides the best performance, yielding a log-likelihood value of 158.93 and a mean absolute percentage error (MAPE) of 0.76%. These findings indicate that the model can accurately represent freight dynamics in Sumatra and may serve as a quantitative reference for regional rail freight planning and infrastructure development.