Development of a NOx Prediction Model for Marine Low-speed Engine

This record contains the dataset and trained models supporting the manuscript entitled **"Development of a NOx Prediction Model for Marine Low-speed Engine"**. The dataset was constructed for NOx prediction under four representative load conditions of a marine low-speed engine: 25%, 50%, 75%, and 100% load. The original computational samples were obtained by combining simulation samples generated using a design of experiments (DOE) approach and representative experimental samples extracted from marine low-speed engine test-bench measurements. The Bayes Bootstrap method was applied to augment the simulation samples under each load condition. This record includes: - original computational datasets;- Bayes Bootstrap augmented datasets;- trained Back Propagation (BP) neural network models;- trained LightGBM models;- normalization files for the BP models;- example scripts for loading the models and performing NOx prediction.