Hepatitis is a dangerous disease because it is a contagious disease and it is not easy to diagnose the disease early. Due to the difficulty of making an early diagnosis, the disease has the potential to become even more severe and increase the mortality rate. Therefore, it is necessary to develop predictive methods that can be used for the early detection of this disease. In this study, a hepatitis prediction method was developed using a random forest (RF) algorithm combined with feature selection using SVM-RFE (recursive feature elimination). Then, because the dataset used does not have a balanced distribution between classes, which is only 20% for the minority class, SMOTE (synthetic minority oversampling technique) is used to deal with this problem. To determine the best parameters in the model, Grid-Search is used as the tuning hyper-parameters. The classifier built with this approach produces 0.879 accuracy, 0.902 precision, and 0.966 ROC performance. This classifier proved to be better than the other classifiers.