Wave height prediction using artificial immune recognition systems (AIRS) and some other data mining techniques

Significant wave height (SWH) prediction is an important task in coastal engineering activities. Due to difficulties and complexities in predicting SWH using traditional numerical and empirical models, in the past decades, a high tendency to use soft computing approaches has been observed. In this paper, an artificial immune recognition system (AIRS), which is a recently developed data mining approach, is utilized for SWH prediction. The results of AIRS model are compared with those of artificial neural networks (ANN), regression tree induction (named M5P), Rough Set Theory, Bayesian networks and support vector regression models. The AIRS has been successfully employed in some engineering applications, but its efficiency and applicability for SWH prediction have not been investigated. In this paper, the AIRS model is utilized for the SWH prediction in the Lake Superior in North America. Comparing the results of this model and five other data mining techniques shows that the AIRS and ANN can outperform other models.