Prediction Method for Ocean Wave Height Based on Stacking Ensemble Learning Model

Wave heights are important factors affecting the safety of maritime navigation. This study proposed a stacking ensemble learning method to improve prediction accuracy wave heights. We analyzed correlation between and other oceanic hydrological features, according eleven such as measurement time, horizontal velocity, temperature, pressure, model inputs. A fusion consisting two layers was established principle learning. The first layer used extreme gradient boosting algorithm, light machine, random forest, adaptive determine deep relations input features. second linear regression fit relation outputs actual heights, using data from four models layer. trained based on 5-fold cross-verification algorithm. paper real test performances model, results showed that mean absolute error squared were at least 35.79% 50.52% better than those models.