Kamsarmax Main-Engine Fuel Consumption Predictor

Browser-based prototype using the exported DataRobot M49 Eureqa Generalized Additive Model. The model predicts main-engine fuel consumption in liters per hour for operating conditions comparable to the Kamsarmax bulk-carrier dataset used in the project.

Operational inputs

Must be greater than zero because the model uses ln(RPM).
Must be greater than zero because the model uses ln(draft).
Predicted main-engine fuel consumption
— L/hr
Scope limitation: This model was developed from a publicly released dataset derived from real operational measurements of one Kamsarmax bulk carrier under the project’s normal-sailing scope. Performance on other vessels, vessel classes, or operating regimes has not been established. Inputs should remain representative of the conditions covered by the training data.

Model validation

The exported formula was checked against the DataRobot scoring JAR on the full validation set and reproduced the same predictions.

Validation RMSE 47.2004 L/hr
Validation MAE 30.4883 L/hr
Validation R² 0.7688
Show exported M49 formula 
Fuel =
923.037576970209
+ 347.571757618796 × Draft
+ 23.6011711401536 × Wind × Swell
+ 47.8622107886719 × RPM × ln(RPM)
+ 3.43080498299164 × RPM × ln(Draft)
+ 5.69826733245991 × Wind × Swell × ln(Draft)
- 215.454656405291 × RPM
- 0.983141922354151 × SOG × STW
- 26.124406310822 × Draft × ln(RPM)
- 78.2891398697926 × Draft × ln(Draft)
- 8.30332085356404 × Wind × Swell × ln(RPM)