Font Size: 

Measuring, reporting and verification of ship fuel consumption are the main requirements imposed by upcoming European regulations. However, the massive amount of navigation data resulting from ship computerization is not easily handled by shipping operators because of the lack of standardized solutions. In this context, modern statistical and machine learning techniques provide effective methods to exploit the massive operational data available on modern ships and, in particular, can be used for building predictive models to estimate fuel consumption. With resort to real operational data collected from a Ro-Pax cruise ship owned by the Italian shipping company Grimaldi Group, this paper presents an extensive comparison study of modern predictive analytical methods (e.g. variable selection, penalized regression, latent variable methods and tree-based ensembles) in order to explore new directions in the analysis of ship fuel consumption.