Maritime traffic has to deal with the risk of collision on a daily basis. Currently, Vessel Traffic Services Operators are provided with short-term prediction methods, used to resolve potential collisions. Research is done to predict collision risk at a larger time horizon, which is expected to provide a Vessel Traffic Services Operator (VTSO) with more time and information to anticipate upon and prevent situations with high risk of collision from developing. This thesis focusses on forming the basis for the prediction component of this goal. The objective is to provide a basic understanding of the process of medium-term behaviour of vessels. This is done by performing a data analysis of a case study of the vessel traffic off the coast of Rotterdam, investigating which variables can be used to predict the intent of a vessel. Two aspects of the intent of a vessel are considered: where the vessel intends to end (within the scope of the scene), and which intermediate waypoints it plans to follow. Entry points, exit points and waypoints are clustered using the Density Based Spatial Clustering of Applications with Noise (DBSCAN) clustering technique. Waypoints are derived by detecting change-points in the course of vessels using binary segmentation. Variables from the dataset are selected and it is investigated which variables can distinguish between different intents, given the entry point of the vessel. The results are that the variables 'course' and 'destination' can distinguish between routes sufficiently enough to investigate them further. This further investigation for the course variable is due to its dependence on vessel position, among other variables. Other variables may add value also, but then in combination with these two variables. The waypoint determination has not yet been successfully implemented, but it is regarded as a promising means to describe and predict vessel intent in more detail, partially due to the conclusions drawn regarding the course variable.