Seaport throughput forecasting by hybrid machine learning algorithm

Port throughput forecasting is not only a complicated problem but also a challenging task in seaport decision support system. This paper deals with novel hybrid model that utilizes discrete wavelet transform (DWT) and the deep learning method - long short-term memory (LSTM) network for port productivity forecasting. Based on the experimental results and related data analytics, it is found that the proposed hybrid method is superior to the five representative models in terms of prediction accuracy and efficiency; thus, advocating for the performance and robustness of the proposed hybrid strategy as well as the suitability in computer-based business decision.