DERN: Deep Ensemble Learning Model for Short- and Long-Term Prediction of Baltic Dry Index

Abstract

The Baltic Dry Index (BDI) is a commonly utilized indicator of global shipping and trade activity. It influences stakeholders’ and ship-owners’ decisions respecting investments, chartering, operational plans, and export and import activities. Accurate prediction of the BDI is very challenging due to its volatility, non-stationarity, and complexity. To help stakeholders and ship-owners make sound short- and long-term maritime business decisions and avoid market risk, we performed short- and long-term predictions of BDI using an ensemble deep-learning approach. In this study, we propose to apply recurrent neural network models for BDI prediction. The state-of-the-art of sequential deep-learning models such as RNN, LSTM, and GRU are employed to predict one- and multi-step-ahead BDI values. In order to increase the accuracy, we assemble the models. In experiments, we compared our results with those of traditional methods such as ARIMA and MLP. The results showed that our proposed method outperforms ARIMA, MLP, RNN, LSTM, and GRU in both short- and long-term prediction of BDI.

Dengue Vector Population Forecasting Using Multisource Earth Observation Products and Recurrent Neural Networks

“…Using the one-week-ahead NARX forecast model proposed in this work, public health managers have more time to plan and respond. Future studies will consider an extension into multi-steps-ahead forecasting, since RNNs have already been successfully applied to such cases in other domains [44], [17]. Additionally, other deep learning network architectures will be considered, such as the transformer model applied in [9] for multi-step-ahead forecasting of influenza epidemics prevalence.…”

A Hybrid LSTM-Based Ensemble Learning Approach for China Coastal Bulk Coal Freight Index Prediction

“…Uyar et al [34] presented one trained recurrent fuzzy neural system with the use of a genetic algorithm for improving long-term dry cargo freight rates prediction to be more accurate. As an efficient strategy to improve the forecasting ability of a single model, ensemble learning has been widely used to improve the model performance [35][36][37][38]. For example, Kamal et al [35] developed a deep ensemble recurrent network of recurrent neural network (RNN), long-short-term memory (LSTM), and gated rectified unit neural network (GRU) to improve the BDI predictive performance, and results showed that the ensemble method outperforms the single deep-learning approach.…”

“…As an efficient strategy to improve the forecasting ability of a single model, ensemble learning has been widely used to improve the model performance [35][36][37][38]. For example, Kamal et al [35] developed a deep ensemble recurrent network of recurrent neural network (RNN), long-short-term memory (LSTM), and gated rectified unit neural network (GRU) to improve the BDI predictive performance, and results showed that the ensemble method outperforms the single deep-learning approach. Tan et al [37] proposed an LSTM-based deep ensemble learning model that combined bagging, random subspace, and boosting to forecast ultra-short-term industrial power demand, and they found out that the proposed model obtains higher accuracy and robustness than LSTM, eXtreme Gradient Boosting (XGBoost), and other time series methods.…”

A Meta-Modeling Power Consumption Forecasting Approach Combining Client Similarity and Causality

“…In the literature, several experiments were conducted with different LSTM variants, including a variable number of units and hidden layers as well as custom training loops for sequence forecasting. However, it is evident that, while changes in the structural parameters of an LSTM can boost model performance and achieve faster training time through faster convergence, stable and reliable results are derived from the aggregation of multiple LSTMs and the construction of ensemble models [34]. Therefore, for the purposes of this study, an LSTM ensemble was considered for the forecasting experiments and the weighted average of the ensemble members was used for each representation of the predicted output.…”