Scholarlink

<p>This study advances the understanding of ocean freight rate forecasting by developing a novel Multi-Modal Analysis and Synthesis (MMAS) framework. Using data from 2001 to 2022, we investigate the complex relationships between uncertainty indices and three major shipping indices: the China Containerized Freight Index (CCFI), the Baltic Dry Index (BDI), and the Baltic Dirty Tanker Index (BDTI). The research employs an innovative methodological approach combining Multivariate Variational Mode Decomposition (MVMD), deep learning models, and SHapley Additive exPlanations (SHAP) analysis. Our findings reveal that the MMAS framework significantly outperforms traditional forecasting methods, particularly during market volatility. The influence of external events varies distinctly across shipping submarkets: container shipping shows primary sensitivity to climate and energy factors; dry bulk shipping responds mainly to energy and economic factors; and tanker shipping demonstrates heightened sensitivity to economic, geopolitical, and climate factors. This paper extends beyond traditional perspectives, showing how sudden external events provide valuable predictive information across multiple temporal scales, thus enhancing our understanding of freight rate dynamics during extraordinary market conditions, such as the COVID-19 pandemic.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An interpretable multi-model ensemble deep learning framework for forecasting ocean freight indices with external uncertainty factors

  • 期刊论文

  • 发表日期：

    2025年5月7日

查看全文
  • Wenyang Wang,
  • Nan He,
  • Peng Shao,
  • Jibin Zhou

摘要

This study advances the understanding of ocean freight rate forecasting by developing a novel Multi-Modal Analysis and Synthesis (MMAS) framework. Using data from 2001 to 2022, we investigate the complex relationships between uncertainty indices and three major shipping indices: the China Containerized Freight Index (CCFI), the Baltic Dry Index (BDI), and the Baltic Dirty Tanker Index (BDTI). The research employs an innovative methodological approach combining Multivariate Variational Mode Decomposition (MVMD), deep learning models, and SHapley Additive exPlanations (SHAP) analysis. Our findings reveal that the MMAS framework significantly outperforms traditional forecasting methods, particularly during market volatility. The influence of external events varies distinctly across shipping submarkets: container shipping shows primary sensitivity to climate and energy factors; dry bulk shipping responds mainly to energy and economic factors; and tanker shipping demonstrates heightened sensitivity to economic, geopolitical, and climate factors. This paper extends beyond traditional perspectives, showing how sudden external events provide valuable predictive information across multiple temporal scales, thus enhancing our understanding of freight rate dynamics during extraordinary market conditions, such as the COVID-19 pandemic.