Enhancing monthly lake levels forecasting using heuristic regression techniques with periodicity data component: application of Lake Michigan

This study investigates the accuracy of three different techniques with periodicity component for estimating monthly lake levels. The are multivariate adaptive regression splines (MARS), least-square support vector (LSSVR), and M5 model tree (M5-tree). Data from Lake Michigan, located in USA, is used analysis. In first stage modeling, were applied to forecast level fluctuations up 8 months ahead time intervals. second stage, influence was (month number year, e.g., 1, 2, 3, …12) as an external subset modeling root-mean-square error, mean absolute coefficient determination evaluating models. both stages, comparison results indicate that MARS generally outperforms LSSVR M5-tree. Further, it has been discovered including input models improves their projecting