A supervised statistical learning approach to the preliminary design of cruise ship Gross Tonnage

Abstract

This work demonstrates how supervised statistical learning methods can be applied to analyse the main characteristics of cruise ships and to settle out a practical formula to support naval architects in the preliminary design of cruise ship Gross Tonnage (GT), which is known to be a conventional measurement of the carrying capacity of the vessel. This application is motivated by the current growing need for optimizing the number and the comfort of onboard passengers. By means of a data set collecting open information on 54 cruise ships delivered after 2000 and still in service, the attained formula has shown excellent performance also with respect to those competitors that do not need information usually unavailable in the preliminary design, such as luxury cabin volume, dead weight, and construction height. The proposed formula may definitely represent a useful up-to-date practical tool to support designers, shipyards, ship owners, and operators that is prone to be generalized with a new data set in the future and stimulate the maritime community to leverage statistical learning approaches to turn data information into value.