Traditional machine learning models for groundwater potential prediction do not consider the optimal combination of factors, which can adversely affect groundwater potential mapping. For this reason, a groundwater potential prediction method based on genetic algorithm optimization support vector machine is proposed. Taking Yiliang County of Yunnan Province as the study area, a total of 15 influencing factors were selected from topography, hydrology, soil, geology, etc.; considering the model performance and the roles of the influencing factors, the optimal factor combinations containing 11 influencing factors were screened using the genetic optimization algorithm; then the groundwater potential prediction model was constructed using the support vector machine method; finally, the model accuracies and areas under the curves of the receiver operating characteristic (AUCs) before and after the optimization of the factors were calculated, and the receiver operating characteristic (ROC) curves of the model and the groundwater potential prediction maps were plotted. The results show that the accuracy of the model before factor optimization is 0.774, the AUC of the validation set is 0.789, the accuracy of the model after factor optimization is 0.777, and the AUC of the validation set is 0.806, which are increased by 0.003 and 0.017 respectively. It can be seen that the accuracy and reliability of the proposed method are superior to the traditional support vector machine method, and the results can provide scientific reference for regional hydrogeological survey and groundwater resource management and planning.