How Interpretable and Trustworthy are GAMs?
This work addresses the problem of inconsistent and potentially misleading GAM interpretations for practitioners in interpretable machine learning, though it is incremental as it builds on existing GAM methods.
The paper investigates the interpretability and trustworthiness of Generalized Additive Models (GAMs) by comparing various training algorithms on real and simulated datasets, finding that high feature sparsity can lead to missed patterns and unfairness, and concluding that tree-based GAMs offer the best balance of sparsity, fidelity, and accuracy.
Generalized additive models (GAMs) have become a leading modelclass for interpretable machine learning. However, there are many algorithms for training GAMs, and these can learn different or even contradictory models, while being equally accurate. Which GAM should we trust? In this paper, we quantitatively and qualitatively investigate a variety of GAM algorithms on real and simulated datasets. We find that GAMs with high feature sparsity (only using afew variables to make predictions) can miss patterns in the data and be unfair to rare subpopulations. Our results suggest that inductive bias plays a crucial role in what interpretable models learn and that tree-based GAMs represent the best balance of sparsity, fidelity and accuracy and thus appear to be the most trustworthy GAM.