Evaporation is a fundamental process in the hydrological cycle. In warm climates, water losses due to evaporation from rivers, canals, and open-water surfaces represent a significant challenge; therefore, the application of reliable predictive models is essential. This study evaluates the performance of three advanced models—Multilayer Perceptron (MLP) neural network, Random Forest (RF) algorithm, and Gaussian Process Regression (GPR)—for predicting evaporation. Six years of daily meteorological data, including mean wind speed, mean air temperature, mean relative humidity, and evaporation, recorded at the Dehkhoda Agro-Industrial meteorological station in Ahvaz, were used. Model predictions were compared against observed values, and all analyses were conducted in MATLAB 2022. Evaluation metrics, including the coefficient of determination (R²) and root mean square error (RMSE), revealed that the MLP model achieved the highest accuracy during the training stage (R² = 0.92, RMSE = 1.44), outperforming the validation (R² = 0.87, RMSE = 1.77) and testing (R² = 0.91, RMSE = 1.68) stages. The overall performance of the MLP model (R² = 0.93, RMSE = 1.40) confirmed its high predictive capability. The overall results for the RF and GPR models were R² = 0.93, RMSE = 1.41, and R² = 0.93, RMSE = 1.42, respectively, indicating that all three models demonstrated high accuracy in predicting evaporation. Based on these findings, the MLP model is recommended as the primary approach for evaporation prediction, followed by RF and GPR