One of the most important factors in the design of fairways for inland waterway traffic is the width vessels require for safe navigation. This is not only given by the constant dimensions of the vessel, but highly influenced by the vessel’s surroundings. Furthermore, it heavily depends on the vessel’s speed being regulated by the steersman. We investigate a complex physical model given as a “black-box,” that describes these dependencies and look for simpler, purely data driven regression based approaches: parametric, non-parametric, and kernel supervised principal component. Some models allow to reduce the dimensionality and identify the most important influencing variables.