Using association rules to map vessel behaviour patterns and port relationships

Abstract

Maritime transportation networks form the backbone of global trade, yet traditional analytical approaches often fail to capture the strategic operational choices that shape port-to-port relationships. This paper introduces a novel methodological framework—Adapted Association Rules Mining (AARM)—to analyse systematic vessel behavioural patterns in maritime networks through bilateral port relationships using comprehensive container vessel movement data from 2016 to 2024. By adapting statistical association metrics to incorporate directionality and frequency, we uncover a hidden strategic architecture within maritime networks. Our analysis of 49,970 relevant port-to-port connections identifies three distinct strategic archetypes: a dominant natural state representing flexible, distributed operations (approximately 80% of connections), and two specialized patterns exhibiting asymmetric dependencies toward either origin or destination ports (the remaining 20%, divided into two similarly sized groups). Geographic analysis reveals that approximately 90% of specialized relationships involve ports separated by 3000 km or less, establishing specialization as predominantly a regional phenomenon. We introduce novel metrics including the confidence asymmetry measure (CAM) and the lift deviation index (LDI) to capture global behavioural structures beyond traditional centrality measures. While these structural patterns remain remarkably stable throughout the entire study period, detailed temporal analysis comparing pre- and post COVID-19 configurations reveals significant heterogeneity in adaptation strategies across regions and vessel size categories. Some regions have reverted to pre-pandemic structures while others underwent permanent strategic reorganization. Through detailed analysis of Singapore’s counter-trend evolution, we demonstrate how individual ports can undergo substantial systematic operational reconfiguration—evidenced by synchronized changes across all AARM metrics—while contributing to overall system stability. Maritime network resilience emerges not from structural rigidity, but from a systematic operational reconfiguration where most connections operate in a non-specialized state while a strategic minority pursues specialized asymmetric relationships, with specialization concentrated at regional scales.