Machine learning models help container terminals reduce wasted moves by predicting service needs and dwell times

Researchers developed ML models using historical operational data to forecast which containers need pre-clearance handling before cargo release

Data preparation included implementing a classification system for cargo descriptions and deduplicating consignee records to improve data quality

The predictive models consistently outperformed rule-based heuristics and random baselines across multiple temporal validation periods

Accurate predictions of container dwell times enable better strategic planning and resource allocation in yard operations