This work presents the application of a data-driven model for streamflow predictions, which can be one of the possibilities for the preventive protection of a population and its property. A new methodology was investigated in which ensemble modeling by data-driven models was applied and in which harmony search was used to optimize the ensemble structure. The diversity of the individual basic learners which form the ensemble is achieved through the application of different learning algorithms. In the proposed ensemble modeling of river flow predictions, powerful algorithms with good performances were used as ensemble constituents (gradient boosting machines, support vector machines, random forests, etc.). The proposed ensemble provides a better degree of precision in the prediction task, which was evaluated as a case study in comparison with the ensemble components, although they were powerful algorithms themselves. For this reason, the proposed methodology could be considered as a potential tool in flood predictions and prediction tasks in general.
Part of the book: Time Series Analysis and Applications