<![CDATA[This research is an endeavour to develop an explainable machine learning framework to quantify the combined effect of ship type, fuel type, distance and weather conditions affecting fuel consumption and CO2 emissions for green maritime operations. The data collected from the voyage-level records for various vessel categories was pre-processed and used to train three supervised regression models: Linear Regression, Random Forest, and Extreme Gradient Boosting (XGBoost). The models were tested based on the coefficient of determination (R2) and mean squared error for training and test data sets separately for fuel consumption and CO2 emission. Results show that all models are able to capture the main trends, but the Random Forest was able to provide the most accurate and robust predictions, with values of test R2 exceeding 0.94 and the lowest values of error for both target variables. In order to improve the interpretability, SHapley Additive exPlanations (SHAP) analysis and feature importance measures were used for the Random Forest models. Distance becomes the main factor, whereas ship type, fuel type, and weather variables have a secondary but significant impact on fuel consumption and emissions. The proposed approach offers a transparent and computationally efficient aid for supporting operational optimization, fuel choice evaluation, and policy design in the context of maritime decarbonization.]]>
Copyrights © 2025
