Understanding shipowner fuel choice decisions in newbuilding orders: An explainable machine learning approach

The transition of the shipping industry toward alternative fuels is a strategic decision, in large part necessitated by international environmental regulations. However, the factors associated with shipowners’ fuel decision at newbuilding stage are not fully understood. This study introduces explainable artificial intelligence (XAI) to transcend traditional analyses, in an effort to reveal and explain the multitude of parameters that play a role in fuel decisions. Through the use of a Light Gradient Boosting Machine (LightGBM) on an extensive newbuilding dataset from Clarksons, interpreting it with XAI, we identify four predictors of fuel choice: vessel size, vessel type, shipowner nationality, and market cycle. Especially the fourth predictor -market cycle-addresses unexpected counter-cyclical investment opportunities in next-generation fuels, such as ammonia. Our findings reconceptualize the fuel-choice dilemma, from a mere technical issue into a multidimensional strategic decision framework. We offer shipowners a data-driven perspective to inform their investment decisions and provide policymakers empirical evidence to design more targeted regulations that account for the diverse realities of the global shipping industry.