Week 11: Other Models — logistic regression, ensembles

Learning Objectives

• 11A I can train and interpret logistic regression models for binary and multiclass classification.
• 11B I can train ensemble models (random forests, gradient boosting) and explain how they improve on a single decision tree.

Perspectival Reading

Reading: TBD

Reflection Questions

1. Ensemble methods sacrifice some transparency for accuracy. When is that tradeoff acceptable, and who gets to decide?
2. Logistic regression outputs a probability — what does it mean to treat that number as a fact about a person?
3. The dominance of certain model families in competitions shapes what gets taught and used. What gets crowded out?

Materials

• Monday class
• Wednesday class