Elucidating US Import Supply Chain Dynamics: A Spatial-Temporal Graph Neural Network Approach

To enhance the understanding of congestion points at ports and provide visibility into the incoming container ships in the USA, this study focuses on maritime ports and corresponding terminals, using several ports along the East Coast of the United States as case studies. We analyzed Automatic Identification System (AIS) data from January 2015 to September 2023, deriving comprehensive his-torical berthing data. This analysis enabled us to build several port congestion prediction models, such as Extreme Gradient Boosting (XGBoost), Long Short-Term Memory (LSTM), and a more sophisticated Spatial-Temporal Graph Neural Network (ST-GNN) model. Additionally, this research traced the historical routes of container ships and accurately mapped the sectors where ships' berth within terminals, thereby providing deeper insights into ship scheduling. This study offers considerable value to stakeholders in the supply chain industry, contributing to both theoretical and practical applications in maritime logistics.