I checked several questions, like Overfitting during model selection - AutoML vs Grid search and Hyperparameter tuning using grid search/randomised search, but I don't think any of them answer my question (even though I believe is a very common one).
So I was using RandomSearchCV from scikit-learn in order to tune the hyperparameters of a Random Forest Regression (and other models). However, I believe I'm getting overfitted models. In some cases, the root mean square error in the training dataset is less than half of the testing dataset. Is there a way of using some sort of Grid Search and not get an overfitted model? I mean, how can I get a good hyperparameter tuning without overfitting my model?