The performance of artificial neural networks in predicting the baltic dry index in moments of crisis

Baltic Dry Index (BDI) is an important factor in dry cargo and container transportation pricing. In this study, the behavior of BDI is analyzed in Covid-19 pandemic and 2008 financial crisis using 1D-CNN, CRNN, LSTM neural network models. The analysis consists of assessment for metric values and conclusion where it's decided which model is better for forecasting of the future values. All of the models trained with 200 epochs and early stopping is applied according to value loss. Each models are trained with Covid-19 and 2008 financial crisis datasets and two different train test data splitting methods. Keywords: Artifical Neural Networks, Deep Learning, Baltic Dry Index, ANN, LSTM, CRNN,1D-CNN, BDI, Forecasting Advisor: Dr. Öğr. Üyesi, Ender Gürgen, İşletme Anabilim Dalı, Mersin Üniversitesi, Mersin