邓必年



摘 要: 孤立点对物流成本预测结果具有干扰作用,而当前模型均没有考虑孤立点的负面影响,预测结果可信度低。为了改善物流成本的预测效果,提出基于剔除孤立点的物流成本预测模型。首先对当前物流成本预测研究现状进行分析,并根据密度方法找到物流成本原始数据中的孤立点,删除这些孤立点,然后对物流成本数据进行聚类,找到物流中隐藏的变化特点,采用最小二乘支持向量机建立物流成本预测模型,最后通过物流成本预测实验对性能进行测试。测试结果表明,该模型消除了孤立点的干扰,提高了物流成本的预测精度,物流成本预测的建模效率得到改善,具有很好的实际应用价值。
关键词: 物流成本; 孤立数据点; 密度方法; 预测模型
中图分类号: TN911.1?34; TP301 文献标识码: A 文章编号: 1004?373X(2017)13?0114?04
Abstract: The isolated point plays the interference effect on the forecast result of the logistics cost, but its negative effect isn′t considered in current model, and the reliability of the prediction result is low. In order to improve the forecasting effect of logistics cost, the logistics cost forecasting model based on isolated point elimination is put forward. The current research status of logistics cost prediction is analyzed. The isolated points in original data of the logistics cost are found out according to the density method, and removed. The data of logistics cost is clustered to seek out the change characteristics hidden in logistics. The least square support vector machine is used to establish the logistics cost prediction model. Its performance is tested by means of the logistics cost prediction experiment. The test results show that the model can eliminate the interference of isolated point, enhance the prediction accuracy of logistics cost, improve the modeling efficiency of logistics cost forecasting, and has perfect practical application value.
Keywords: logistics cost; isolated data point; density method; prediction model
0 引 言
随着经济、人们生活水平以及交通技术等不断改善,物流企业越来越多,导致企业之间的竞争加剧,物流成本预测直接影响物流企业的经济效益,是物流研究领域中的一个重要方向,引起了人们的高度关注[1]。
物流成本与一个地区的经济、政策以及交通状况密切相关,是一个复杂多变系统,最原始的物流成本预测通过手工方式实现,一些专业人员采用统计学理论对物流成本进行分析和预测,该方式对小规模物流成本预测可以实现,对于现代大规模物流成本计算过程太复杂,工作效率低,而且易出错。随后有学者提出了物流成本自动预测模型。……