The “Two Cultures”

y = βX + ϵ

• Inference (β)
• Prediction (y)

Modeling

Regression  → f(X) = α + βX + ϵ

CART  → f(X)= Tree

Random Forest  → f(X)= Tree Ensemble

Gradient boost  → f(X)= Sequence of predictors (e.g. trees)

This block

• Baseline model (β + y)
• Fitting Random Forests
• Predictive checking (β + y)
• Uncertainty (β)
• Model performance (y)
• Overfitting & Cross-Validation (y)