Forecasting Container Shipping Prices Under the Influence of Major Events

Irregular events like natural disasters can cause drastic fluctuations in container ocean freight prices, making it challenging for traditional forecasting techniques to accurately forecast their complex dynamics. The ensemble empirical mode decomposition (EEMD) method and the gated recurrent unit (GRU) network are suitable for the cases. Therefore, this paper proposes an EEMD-GRU combined forecasting model. Empirical analysis is conducted using the China containerized freight index (CCFI) data from March 2020 to March 2022. The forecasts of the EEMD-GRU model were compared with those of the ARIMA, LSTM, GRU, EEMD-ARIMA, and EEMD-LSTM models. The results indicate that the proposed significantly outperforms its rivals in terms of MAPE, RMSE, and MAE. This shows good potential of EEMD-GRU to be a powerful tool for stakeholders and government authorities.