Integration of Machine Learning and DEA to Evaluate Efficiency of Adriatic Container Ports

In this study, a hybrid approach combining Machine Learning (ML) with Data Envelopment Analysis (DEA) is proposed to evaluate the operational efficiency of container ports using Automatic Identification System (AIS) data. Representative AIS data on vessel calls were collected for 10 Adriatic container ports in 2025. First, key operational indicators such as vessel turnaround time, berth utilization and cargo throughput were extracted. ML techniques were then applied to improve data completeness and group the ports based on operational similarities. DEA was subsequently used to evaluate the efficiency of ports both globally and within port clusters. The results show that ML models can effectively estimate missing outputs and clustering improves comparability between ports. Our random forest regression model achieved an R2 of 0.89 in predicting cargo volume, supporting more complete and reliable DEA assessments. This cluster-based DEA analysis reveals meaningful performance differences even between operationally similar ports and provides more targeted benchmarking insights. This research advances previous studies on DEA efficiency assessment of ports by incorporating predictive analytics and unsupervised learning into DEA efficiency assessment of container ports.