Time-Aware Validation of Machine Learning Fuel Consumption Models: Evidence from 1\,Hz Operational Data, CCGS \textit{Sir Wilfrid Laurier}
2026-08-17 • Machine Learning
Machine Learning
AI summaryⓘ
The authors looked at how ship fuel use is predicted and noticed that many past studies tested their models in ways that accidentally let future information leak into the training, making results look better than they really are. They used special methods that respect the order of data over time to get a more realistic check of these models. They tested different models using data from a Canadian coast guard ship and found that this time-aware testing gives a clearer picture of how well the models work in real-world situations.
ship fuel consumptiontime series cross-validationtemporal leakageregression modelssteady-state datablocked cross-validationmachine learning validationmaritime transportationmodel evaluation
Authors
Samarasimha Reddy Chittamuru, Ayhan Akinturk, Allison Kennedy, Joshua Barnes, Matthew Hamilton
Abstract
Ship fuel consumption (SFC) prediction supports vessel operation optimisation, emissions estimation, and decision support systems (DSS) for sustainable maritime transportation. Numerous data-driven fuel models have been developed over the past two decades, but a critical and often overlooked limitation lies in their validation practices: most studies evaluate performance using random train--test splits, which, applied to high-frequency records, admit temporal leakage and yield optimistic results that do not reflect deployment conditions. This paper examines that gap using time-aware evaluation, specifically Time Series Cross-Validation (TSCV) and Blocked TSCV (BTSCV). Using the Canadian Coast Guard Ship (CCGS) \textit{Sir Wilfrid Laurier} as a case study, six regression models and a physics baseline are tuned under three time-aware schemes and three feature configurations, then evaluated on a common chronological hold-out set drawn from approximately 3.88 million steady-state 1\,Hz records.