Wave overtopping predictions using an advanced machine learning technique

Coastal structures are often designed to a maximum allowable wave overtopping discharge, hence accurate prediction of the amount is an important issue. Both empirical formulae and neural networks among commonly used tools. In this work, new model for mean discharge presented using innovative machine learning technique XGBoost. The selection features train on carefully substantiated, including redefinition existing obtain better performance. Confidence intervals derived by tuning hyperparameters applying bootstrap resampling. quality tested against four physical data sets, thorough quantitative comparison with methods presented. XGBoost generally outperforms other test sets normally incident waves. All data-driven show less accuracy oblique data, presumably because these conditions underrepresented in training data. performance significantly improved adding randomly selected part cases end, shown reduce errors all work factor up 5 compared methods.