Research on Logistics Demand Forecast of Port Based on Combined Model

Abstract

In order to improve the accuracy, practicability and prediction accuracy of the single prediction model, a logistics demand combination forecasting model based on cargo throughput is established. Based on the original statistical data of throughput, the two models of gray and exponential smoothing are established respectively. On this basis, the weighted assignment method of variance reciprocal is used to construct the combined forecasting model. According to the prediction results of different prediction models analyzed by the three evaluation indexes of average relative error, maximum fitting error and minimum fitting error, the three evaluation indexes of the combined prediction model are respectively 6.344%, 16.345% and 0.343%, which are smaller than the single item model. It indicates that the established combined forecasting model can effectively improve the accuracy of the throughput prediction model based on overcoming the shortcomings of the single-term throughput prediction model.