Port Throughput Forecasting Based On Combination Forecasting Model

As an important node in the allocation of production factors and integrated transport system,the development level of ports has a significant impact on promoting social and economic growth and the construction of integrated transport network.As an important index to measure the level and capacity of port development,the correct prediction of port throughput is of great significance for port scientific development strategy and reasonable planning and layout.With the comprehensive development of the Belt and Road Initiative and the construction of a transportation power,higher requirements are put forward for the construction of inland water transport system.Based on this background,Hengyang Port is carrying out a new round of port planning.This paper takes the port throughput as the research object,aiming to explore a reasonable and accurate port throughput prediction model,so as to provide some reference for the port planning of Hengyang Port in the new period.First of all,based on the domestic and foreign research on the influencing factors of port throughput,this paper studies and analyzes the influencing factors of port throughput from three aspects:the economic situation of port hinterland,the collection and distribution capacity of port hinterland and the conditions of port itself.Combined with the actual situation,15 quantifiable indexes affecting the throughput of Hengyang port are preliminarily selected.On this basis,the primary indexes are screened by grey correlation analysis and principal component analysis,and the key influencing factors of Hengyang port throughput are further obtained,which lays a data foundation for the prediction of Hengyang port throughput.Secondly,when predicting the port throughput of Hengyang port,the multivariate linear regression model and grey GM(1,1)model are constructed respectively,and the prediction results of the two single models are tested.Among them,when using the grey GM(1,1)model to predict,the error is large,mainly because the actual data of Hengyang Port’s throughput are volatile.Therefore,the Markov chain model is introduced to correct the prediction results of GM(1,1).At the same time,the accuracy of the two models is tested through the posterior error ratio and small probability error.The results show that the grey-Markov model can better reflect the change of Hengyang Port’s throughput.Finally,based on the two prediction models,this paper constructs a combined prediction model by determining the weight through the reciprocal weighting method of variance.According to the five indicators of mean square error,mean absolute error,square sum error,mean percentage error and mean relative error,and combined with the fitting diagram of prediction results,the three prediction models are evaluated.The results show that the accuracy of the combined prediction model is higher than that of each single model,and the error indicators are optimal.The fitting of the prediction curve with the actual value curve is excellent,which is suitable for the prediction of port throughput of Hengyang Port.At the same time,the port throughput of Hengyang Port in the next five years is predicted on this basis.It provides a certain reference for the new round of port planning of Hengyang Port.