Evaluation of Different Forecasting Methods for the Estimation of Cargo Throughput in Manila International Container Terminal, Philippines

Abstract

Philippine maritime transport serves as an important corridor for international and domestic trade. In order to support the growing demand for trade, port authorities have been under pressure to improve port efficiency and make services more competitive to accommodate future demand. This paper studies the performance of different forecasting methods for the cargo throughput demand of the Manila International Container Terminal (MICT). Three forecasting methods were implemented and compared, namely Autoregressive Integrated Moving-Average model (ARIMA), Artificial Neural Network (ANN) model, and Vector Autoregression (VAR) model. Macro-economic variables were used as exploratory variables. The predicted results were compared using the coefficient of determination (R²), and root mean squared error (RMSE). Results show that all three forecasting models exhibit goodness of fit and accuracy of results according to their R² and RMSE values. In comparison, the VAR model is the most accurate among the three followed by the ANN and ARIMA models, respectively.