Unit 1

Interpretable Machine Learning

Build models whose prediction process can be inspected directly, from regression coefficients and decision rules to prototype and monotonic neural networks.

Chapter 1

Regression

Build interpretable models from weighted sums, then learn where assumptions, scale, sparsity, and nonlinear outcomes complicate the story.

1. Linear Regression→
2. Sparse Models→
3. Logistic Regression→
4. Generalized Models: GLMs and GAMs→

Chapter 2

Trees and Rules

In this chapter, we represent decisions as paths and rules and discuss the tradeoffs between complexity and accuracy.

1. Decision Trees→
2. Sparse Decision Trees (GOSDT)→
3. Decision Rules→
4. RuleFit→

Chapter 3

Neural Networks

Neural Networks have a reputation for not being inherently interpretable. We document a few attempts to make them more interpretable through the addition of specific layers.

1. Prototype-Based Neural Networks (ProtoPNet)→
2. Monotonic Neural Networks (MonoNet)→
3. Kolmogorov-Arnold Networks (KAN)→