A novel hybrid approach to Baltic Dry Index forecasting based on a combined dynamic fluctuation network and artificial intelligence method

X. Zhang, M.Y. Chen, M.G. Wang, Y.E. Ge and H.E. Stanley

Applied Mathematics and Computation, 2019, vol. 361, issue C, 499-516

Abstract: Enhancing the accuracy of freight rate forecasting is crucial when agents in the shipping industry make their business decisions and strive to avoid or reduce possible risks. Although there has been a lot of freight rate trend analysis in the literature, the time-varying and volatile nature of the shipping market makes the accurate prediction of it extremely difficult, if not impossible. In this research, we first propose the use of a dynamic fluctuation network (DFN) to transform a time series data set into an evolving directed network, which removes the noise in the data and allows us to extract time-varying features in it. We then develop a hybrid approach that combines the DFN and artificial intelligence (AI) techniques to forecast the Baltic Dry Index (BDI), which is new to the literature. The utilization of DFN with AI enables the non-linear, cyclical and dynamic features of the BDI to be extracted effectively and the prediction accuracy is not impacted by the length and time-scale of sample selection either in long-term or short-term forecasting. These advantages of DFN offset the well-known limitations of traditional AI-based algorithms and econometric models in BDI forecasting. The empirical results from applying the resultant model to multi-time-scale datasets in a random sampling case show that the model is more accurate than the model based on the AI technique only. We test the accuracy of the DFN-AI model in three challenging cases which respectively contain a sudden rise, a decline, or frequent fluctuations of the BDI. The DFN-AI model has fewer errors and a higher trend matching rate than the corresponding AI-based model. Our hybrid approach also shows its superiority in working with data containing an extreme market downturn, which shed light on the predictability of BDI. This study has important implications for overall business, commercial, and hedging strategies in the shipping industry.

Keywords: Artificial intelligence algorithm; Complex network; Dynamic fluctuation network; Baltic Dry Index prediction (search for similar items in EconPapers)
Date: 2019
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (10)

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S009630031930445X
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:apmaco:v:361:y:2019:i:c:p:499-516

DOI: 10.1016/j.amc.2019.05.043

Access Statistics for this article

Applied Mathematics and Computation is currently edited by Theodore Simos

More articles in Applied Mathematics and Computation from Elsevier
Bibliographic data for series maintained by Catherine Liu ().