Ensemble modeling approach for rainfall/groundwater balancing

This paper introduces an application to real data of the bias/variance trade-off theory, which has been intensively stud-ied in the fields of neural networks and machine learning. It provides further empirical evidences that genetic program-ming can greatly benefit from this approach in forecasting and simulating physical phenomena. Considerations on the contribution of bias and variance to the total error, and ensemble methods to reduce errors due to variance, have been tackled together with specific application of ensemble modeling to hydrological forecasts. This work copes with a sim-ple averaging method for obtaining more reliable approximations using unaltered symbolic regression. Further consid-erations have been taken into account, such as the influence of GP parameters settings on model’s performances.