基于K-近邻核估计的港口空箱吞吐量预测研究

Forecast of Empty Container Throughput Based on K-NN Kernel Estimation

空箱吞吐量的预测结果可以作为空箱调运的参考,有利于港口的合理规划和作业安排。以香港港口为例,采用K-近邻核估计对港口空箱吞吐量进行回归计算,将计算结果与参数回归方法的计算结果进行比较,表明K-近邻核估计的拟合效果和预测精度都优于多元线性回归方法的拟合效果和预测精度。

Forecast of empty container throughput can not only provide reference for dispatch and allocation of empty containers, but also benefit the reasonable planning and task assignment of a port. In this paper K-Nearest Neighbors Kernel Estimation method was applied to forecasting the throughput of empty containers at Hong Kong Port based on regression analysis, which was compared with parametric regression. The results show that the fitting effect and prediction accuracy of K-Nearest Neighbors Kernel Estimation method is much better than multiple linear regression.