The adoption of indirect test for analog and RF integrated circuits (ICs) can tackle the rising costs of the classical industrial testing of these circuits, hence relaxing the requirements on test equipment. Based on machine-learning techniques, the concept of indirect test is to create a mapping between an indirect and low-cost measurement and the performance of the circuit by training a regression model. In this work, we explore the potential benefit of using ensemble learning. Instead of using a single regression model to predict the performance, the use of ensemble learning consists of combining multiple regression models to enhance the model's generalization. Different ensemble methods based on bagging, boosting or stacking are investigated and compared to classical individual models. Results are illustrated and discussed on three RF performances of a LNA for which we have production test data.