<p>The assessment of groundwater quality is an important issue in water management in developing countries, as groundwater plays a crucial role as a source of drinking water in rural areas. In this study, the suitability of groundwater quality was investigated using the groundwater quality index (WQI). Traditionally, the calculation of WQI is time consuming and often associated with various errors in the calculation of sub-indices. To overcome these challenges, a hybrid model based on machine learning is developed in this study to predict the WQI in a central Mediterranean region in southern Turkey. The Grey Wolf Optimization (GWO) and Particle Swarm Optimization (PSO) algorithms were used to optimise the parameters of the Support Vector Regression (SVR) models used to predict the WQI. The data was collected at 50 different locations in July 2022 during a period of intense irrigation activity. The performance of the models was evaluated using a number of statistical indicators, a number of visual graphics such as Taylor diagrams, scatter plots and violin plots. Four different input combinations were compared and the performance of hybrid models increased as the number of inputs increased. However, given the increased complexity of the model with the addition of each input and the associated difficulties in data collection, it is recommended that Model 2 be used. The SVR-PSO2 model demonstrated an optimal alignment with the observed data, as evidenced by the statistical indices R² = 0.968, RMSE = 2.160, WI = 0.992, and KGE = 0.982. The investigation further revealed that the proposed hybrid models markedly diminish the uncertainty of the model in comparison to the conventional SVR model. The results of the study demonstrate the potential of hybrid modelling techniques for predicting groundwater quality and the crucial role that these models play in the management and protection of water resources.</p>