Generalized Additive Models: Explainability Strikes Back
summary
Claims concrete wins: GAM variants match or beat black‑box models on tabular data, show up in Kaggle winners and a study of 68,000 models, and a case where a seasonal+trend GAM outperforms “lag lma” — 42% more accurate, 1,000× faster. Implementation uses spline bases and penalization with libraries like pyan and pyro to control wiggliness, produce posterior uncertainty, and decompose effects for straightforward, auditable explanations.
AI-written summary · about
chapters (6)
- 0:00Introduction to GAMs
- 13:13What are GAMs
- 17:50Engel curve example
- 27:08Real-world applications
- 30:20GAMs limitations
- 31:27Q&A
related talks
32mpyGAM: balancing interpretability and predictive power using...
43mToward Human-Centered Machine Learning
28mExplaining Machine Learning Models with LLMs
28mManaging AI Risk: Beyond Explainability
27mInterpretability and explainability in machine learning
26mStop Explaining Black Box Models, Use Interpretable Models Instead
01h 01mINTERPRETABLE AND DIFFERENTIALLY PRIVATE MACHINE LEARNING
45mExplainable Machine Learning Explained