Part XIV — Model Selection

Choosing among candidate models without fooling yourself: the bias–variance tradeoff, cross-validation, information criteria, feature selection, and model averaging.

Topics

Topic	Focus
Bias–Variance Tradeoff	Underfitting versus overfitting in expected error
Cross Validation	Estimating out-of-sample error by data splitting
Information Criteria (AIC/BIC)	Penalized-likelihood scores for comparing models
Feature Selection	Choosing predictors without contaminating inference
Model Averaging	Combining models instead of picking one