The Application of SARIMAX and LSTM Models under Bayesian Optimization in Cargo Throughput Forecasting of Rizhao Port
In global trade activities,accurately predicting port cargo throughput is significant for logistics and supply chain management.To better capture the seasonal cycles and long-term trends in throughput data,this study employs the SARIMAX model and LSTM neural network,while also considering macroeconomic indicators such as GDP and total import and export values.This paper first utilizes the SARIMAX model to capture the seasonal and trend components in the throughput data,combined with Bayesian optimization for meticulous adjustment of the model's hyperparameters.Subsequently,it introduces an LSTM network integrated with Bayesian optimization to correct the error sequence between the SARIMAX model's predicted values and actual values.Empirical results indicate that,compared to traditional single models,a combined model considering various influencing factors can effectively improve the accuracy of future cargo throughput forecasting.This study not only enriches the theory of time series forecasting but also provides a practical decision-support tool for the port and logistics industries.
time series forecastingSARIMAXLSTMBayesian optimizationcargo throughput
NETL
NSTL
万方数据
