Driver anomaly detection in cargo terminal

The Iranian road transportation sector, comprising about 500,000 owner-operator drivers, faces rising syndication challenges, leading to disruptions and driver refusals in some provinces. Drivers highlight the urgent need for load distribution improvements within terminals. This study investigates anomaly detection by drivers in cargo terminals, starting with the evaluation of driver assumptions through K-means clustering. The study confirms drivers' assertions regarding those who handle more cargo in less waiting time. Subsequently, Isolation Forest, KNN, and HBOS algorithms are applied to detect abnormal behavior using data mining techniques. Results reveal three distinct driver groups, with a notable proportion (98 %) of anomalies concentrated in one group. This study sheds light on the critical syndication issue of anomaly detection in cargo terminals by drivers, offering valuable insights for researchers and shipping practitioners. Moreover, limited research on theft prevention renders conventional methods ineffective, highlighting the overlooked use of clustering in prior literature reviews focused on case study analysis.