Evaporation Prediction with Wavelet-Based Hyperparameter Optimized K-Nearest Neighbors and Extreme Gradient Boosting Algorithms in a Semi-Arid Environment

The study aims to reveal which mother wavelet type performs best in evaporation prediction. This study used a hybrid algorithm that combined K-Nearest Neighbors (KNN), Extreme Gradient Boosting (XGBoost) methods, and discrete wavelet transform to estimate the evaporation values ​​in the summer months in Bursa. While establishing wavelet-machine learning models, Daubechies2, coiflet 2, symlet 2 and biortogonal 2.2 wavelets were obtained by separating various input variables into sub-signals. For the model’s setup, total precipitation, average temperature, average relative humidity, average wind speed, total sunshine duration values in the monthly time period were used as inputs. Total evaporation data were presented as output of the algorithms. The performances of the established models were evaluated with the coefficients of determination, root mean square error, mean absolute error and scatter diagrams. It has been determined that the optimum model is obtained with the hybrid algorithm established by the combination of biorthogonal 2.2 and KNN methods. The study results provide agricultural, meteorological, hydrological and environmental engineers with essential resources for planning and managing water-related issues. © 2023, The Author(s), under exclusive licence to Springer Nature Switzerland AG.