This paper presents a machine learning-based approach to predicting vessel arrival times and optimizing berth allocation, two critical challenges in maritime logistics. Accurate arrival time predictions are essential for efficient port operations, minimizing congestion, and reducing logistical pressures on supply chain stakeholders. In this study, historical data from the Norwegian Base Station and satellite observations, collected between August 1 and September 24, 2024, were used to develop predictive models. After comprehensive preprocessing, three regression models—Gradient Boosting, K-Nearest Neighbors (KNN), and Random Forest—were evaluated for their predictive accuracy. Random Forest achieved the highest performance, with an $R^2$ score of 0.704 and a Mean Absolute Percentage Error (MAPE) of 0.0285\%, demonstrating superior generalization and predictive power. These results underscore the effectiveness of machine learning in enhancing vessel arrival time predictions, leading to more efficient berth scheduling and improved port operations. |
*** Title, author list and abstract as submitted during Camera-Ready version delivery. Small changes that may have occurred during processing by Springer may not appear in this window.