科研管理 ›› 2018, Vol. 39 ›› Issue (10): 148-158.

• 论文 • 上一篇    下一篇

基于VMD-WA-RFR模型的BDI指数影响因素研究

武华华1,2,匡海波1,张鹏飞1,2   

  1. 1. 大连海事大学综合交通运输协同创新中心, 大连116026;
    2. 大连海事大学交通运输工程学院, 大连116026
  • 收稿日期:2018-01-17 修回日期:2018-09-13 出版日期:2018-10-20 发布日期:2018-10-12
  • 通讯作者: 匡海波
  • 基金资助:

    国家自然科学基金重点项目(71831002),时间:2019-2023;国家自然科学基金 (71672016),时间:2017-2020;教育部长江学者和创新团队发展计划(IRT_17R13),时间:2018-2020;大连海事大学综合交通运输协同创新中心种子基金(3132018301,3132018304),时间:2018-2019;辽宁省高等教育内涵发展专项资金资助项目(20110117406),时间:2017-2018。

A research on the influencing factors of BDI based on the VMD-WA-RFR model

Wu Huahua1,2, Kuang Haibo1, Zhang Pengfei1,2   

  1. 1. Collaborative Innovation Center for Transport Studies, Dalian Maritime University, Dalian 116026, Liaoning, China;
    2. Transportation Engineering College, Dalian Maritime University, Dalian 116026, Liaoning, China
  • Received:2018-01-17 Revised:2018-09-13 Online:2018-10-20 Published:2018-10-12

摘要: 本文全面考虑BDI指数不同类型的影响因素,将影响因素划分为供给方面、需求方面以及宏观经济三种类型,采用基于VMD分解、小波分析与随机森林回归算法构建的VMD-WA-RF模型,综合分析各影响因素与BDI指数的关系,增强了BDI指数影响因素分析的数学理论。进一步的研究发现:一是,借助VMD-WA-RF模型能够得出BDI指数影响因素的重要性排名,有助于有针对性的调节航运市场;二是,能够得出最重要的五个影响因素分别对BDI指数的影响程度,科学地找出能够迅速调节当前航运市场最有效的方案;三是,通过随机森林回归模型,对BDI指数进行预测,表明本模型预测的精确度较高。

关键词: BDI指数, 随机森林, 重要性, 预测

Abstract: This paper considers the influence factors of BDI index of different types, the factors can be divided into supply and demand and macroeconomic three types. Using the VMD-WA-RF model is constructed based on VMD decomposition, wavelet analysis and random forest regression algorithm, to study the comprehensive relationship between the influence factors and the BDI index. It enhanced the mathematical theory of the analysis of the factors affecting the BDI index. A further study found that: First, the VMD-WA-RF model can get the importance ranking of the factors affecting the BDI index, which contributes to the targeted regulation of the shipping market; Second, it can draw the degree of the most important five factors that affect the BDI index respectively, and find out the most effective way to adjust the current shipping market scientifically; Third, the BDI index is predicted by the random forest regression model, which shows that the accuracy of this model is higher.

Key words:  BDI index, random forest, importance, forecast