Port Type Prediction Based on Machine Learning and AIS Data Analysis

The estimated number of main ports and stationary areas in the world account for almost 25,000 and most of them are not well known. Being able to provide the navigator with information such as the type of surrounding ports in his navigation area is therefore of interest. Automatic Identification System data transmitted by ships is a valuable source of information whose potential can be exploited to give further knowledge on the maritime situation. It is also useful to extract knowledge about ports’ activities and types. This paper aims at analysing AIS data using machine learning methods, and more specifically supervised classification to establish a harbour map with the objective of identify port’s type especially for the less documented areas of the globe.