Container port performance assessment: a nonnegative matrix factorization approach

Maritime transport is the backbone of globalized trade and the manufacturing supply chain. Container ports, as a result, have become critical nodes in global supply chains. How a container port performs is therefore a crucial element in a country’s cost of trade. Constructing a reliable, consistent, and comparable basis on which to compare operational performance across different ports is a major challenge. In this work, we leverage automatic identification system data and time stamps from vessels to construct a data-driven ranking of container ports. The focus is purely on quayside performance to be reflective of the experience of a ship operator, the port’s main customer. As we show, the average port time of a ship can be well-approximated using a small number of latent factors. We propose a new nonnegative matrix factorization algorithm to estimate these factors and produce an index that can be used to rank the ports' performance. In general terms, the port with the highest ranking has the least total port time. We illustrate how the new approach is competitive compared to other natural ranking methods using numerical simulations.