Prediction of international shipping container throughput based on particle swarm optimisation and grey wolf optimisation

In order to improve the prediction accuracy for port container throughput in international shipping, PCA is first used to reduce the dimension of the input SVR indicators, thereby reducing the dimension of the SVR input. Secondly, GWO is used to improve PSO, and a port container throughput prediction model based on GWO-PSO-SVR is constructed, thereby improving the prediction accuracy of SVR for port container throughput. Results show that the improved PSO performs well in the test function. Based on data from Tianjin Port, the SVR prediction results indicate that its MAPE index is the lowest, at 12.96, which is closest to the true value.