Comparing models to forecast cargo volume at port terminals

Economic growth has a direct link with the volume of cargo at port terminals. To encourage growth, investment decisions on infrastructure are required that can be performed by development econometric models. We compare three time-series models and one machine-learning model to estimate forecast volume. apply an ARIMA+GARCH+Bootstrap, multiplicative Holt-Winters, support vector regression model, explanatory variables ARIMAX. The through ports San Pedro using data from 2008 2016. database contains imports exports bulk, container, reefer, ro-ro cargo. Results show Holt-Winters is best method bulk cargo, while Diebold-Mariano Test, RMSE metric, MAPE metric validate results.