Research pick: Reshuffling shipping - "Container transaction type prediction: a seaport case in Turkey"

A study of a container shipping terminal in Turkey, published in the International Journal of Shipping and Transport Logistics, demonstrates how it is possible to predict with almost 90 percent accuracy the transaction types that will be needed once an inbound container vessel docks at the quayside before it arrives and so reduce the need for logistics planners to shuffle containers unnecessarily.

Container reshuffling is a necessary evil at container terminals the world over. The approach attempts to solve the problem of uncertainty surrounding the transactions that will need to be carried out on hundreds and thousands of incoming containers arriving from distant ports. The problem is exacerbated by changes that occur in cargo ownership in transit, details going missing overseas, and other disruptions that mean inventory and manifest may not match the logistics planned by the container port for the next vessel.

Given that the new generation of supersized container ships can carry almost 25000 containers, it is obvious that reshuffling is a big issue for a busy port where several container ships may be docked within a relatively short period of time and all require unloading of inventory in as timely and efficient a manner as possible. Reshuffling can wreak havoc on port efficiency, leading to delays, operational inefficiencies, and even lost containers.

Elifcan Dursun and Sule Gungor of Tarsus University in Tarsus, Mersin, Turkey turned to the Cross Industry Standard Process for Data Mining (CRISP-DM) framework. They have used its strategic approach to develop a predictive model that could preclude the need for container reshuffling by allowing the planners and logistics managers to be almost wholly confident in their allocation of inbound containers.

While a Turkish port was used as a case study, the approach could be a guiding light for other ports grappling with similar uncertainties regarding transaction types and relying on container reshuffling as their modus operandi. By harnessing data mining techniques to predict transaction types, the new predictive model all but eliminates the disruptions that are normally caused at ports by the need for container reshuffling.

Dursun, E. and Gungor, S. (2023) ‘Container transaction type prediction: a seaport case in Turkey’, Int. J. Shipping and Transport Logistics, Vol. 17, Nos. 1/2, pp.41–59.