A new exploration in Baltic Dry Index forecasting learning: application of a deep ensemble model

Abstract

World trade is growing constantly, facilitated by the fast expansion of logistics. However, risks and uncertainty in shipping have also increased, in dire need to be addressed by the research community, through more accurate and efficient methods of forecasting. In recent years, combining attention models and deep learning has produced remarkable results in various domains. With daily data spanning the period from January 6, 1995, to September 16, 2022 (totaling 6896 observations), we predict the Baltic Dry Index (BDI) using a deep integrated model (CNN-BiLSTM-AM) comprising a convolutional neural network (CNN), bi-directional long short-term memory (BiLSTM), and the attention mechanism (AM). Our findings indicate that the integrated model CNN-BiLSTM-AM encompasses the nonlinear and nonstationary characteristics of the shipping industry, and it has a greater prediction accuracy than any single model, with an R 2 value of 96.9%. This research shows that focusing on the data’s value has a particular appeal in the intelligence era. The study enhances the integrated research of machine learning in the shipping business and offers a foundation for economic decisions. © 2023, The Author(s), under exclusive licence to Springer Nature Limited.