Fusing Birch with G. Boosting for improving temporal traffic congestion tailored to port gates: Case Study in Patras, Greece

Abstract

Traffic congestion is a vulnerable aspect of every port producing inconvenient situations for all stakeholders. This paper suggests a model that simulates traffic density at ports' gates into three levels: low, medium and high. Parameters that affect traffic density are clustered to strengthen potential relationship between them by using BIRCH algorithm. Furthermore, Gradient Boosting Algorithm is utilized to classify traffic volume into the three states while exploiting available historical data along with the clustered data. The aim is to propose a non-intrusive and inexpensive novel tool for traffic simulation based only on available port data while ensuring better quality of transportation particularly with regard to boarding time towards avoiding further delays. Finally, the proposed traffic simulation model is evaluated with real-time port data, while the experimental results have shown that it is a powerful tool for port traffic simulation and management.