Research On The Throughput Prediction Of China’s Foreign Trade Port Under The COVID-19 Epidemic Situation

China’s foreign trade ports are not only the embodiment of the construction level of urban transportation infrastructure,but also an important part of supporting international economic and trade exchanges and cooperation.The production and operation activities of the port have played a vital role in promoting regional economic development and overall social development.Port throughput is an important index to measure the operation effectiveness of port enterprises.Scientific prediction,analysis and Research on it will help enterprises adjust their production and operation plans in time to adapt to the actual external environment.The outbreak of the novel coronavirus pneumonia in January 2020 has rapidly impacted the economic and social development of all countries in the world,and has greatly increased the uncertainty of the global trade market.Therefore,in order to improve the effectiveness and accuracy of China’s foreign trade port throughput prediction,we can add domestic and foreign epidemic related factors for analysis.Firstly,according to China’s foreign trade port throughput,taking the throughput data of Guangzhou port group over the years as an example,this paper decomposes the long-term,seasonal,fluctuating and irregular trend characteristics of the throughput data,constructs the SARIMA time series prediction model,and analyzes the results,It is proposed that the prediction of port throughput under the epidemic situation should be combined with the actual external environmental factors,and the effects of domestic and foreign epidemic development,hinterland economic and trade level,port collection and distribution capacity and other external factors on port throughput are analyzed.Secondly,combined with the analysis of influencing factors,a preliminary prediction index system is established,and the key indexes that can be used for actual prediction are extracted by stepwise regression method,including the number of newly diagnosed domestic epidemic cases,the number of cured epidemic cases in the province,the number of deaths from foreign epidemic cases,the total value of commodity import and export,the total urban cargo volume,the turnover of waterway cargo and the historical throughput of the port.Furthermore,this paper constructs a port throughput prediction model under epidemic situation based on improved support vector regression,and introduces particle swarm optimization(PSO)and recursive feature elimination algorithm(RFECV)to adjust the parameters and determine the characteristics of support vector regression prediction model(SVR),which verifies the effectiveness of PSO-SVR-RFECV model.Compared with the ordinary multiple linear regression model,it has higher accuracy and better stability.Finally,because the port throughput data under the epidemic belongs to the prediction range of small samples,for the generalization and reliability of prediction model,this paper cross verifies the port throughput prediction model,and puts forward a cross platform interaction framework based on PMML combined with the actual use environment of Guangzhou port production and business system.At the same time,according to the contribution analysis of the influencing factors of the prediction model,the corresponding management suggestions for the coordinated development of the port and the hinterland are put forward.