Tree-based models' feature attributions can vary based on the method used, impacting interpretability.