Topic: Information Criteria and Model Comparison

Worked Mallows Cp Model Comparison

Worked AIC Ranking for Competing Regressions

Worked BIC Comparisons Across Sample Sizes

Why AICc Adds a Small-Sample Correction

Why Ordinary R-Squared Cannot Rank Every Model

The Fit-Complexity Tradeoff

Near-Ties and Model Uncertainty

Mallows Cp for Least-Squares Models

Deriving AIC’s Two-Parameter Penalty

Deriving BIC’s Log-Sample Penalty