Miss World 2004

Model Selection

Miss Universe 2003

Introduction to Model Selection (.pdf)


Important Publications


'Model selection (variable selection in regression is a special case) is a bias versus variance trade-off and this is the statistical principle of parsimony. Inference under models with too few parameters (variables) can be biased, while with models having too many parameters (variables) there may be poor precision or identification of effects that are, in fact, spurious. These considerations call for a balance between under- and over-fitted models -- the so-called "model selection problem" (see Forster 2000, 2001).'
Burnham and Anderson (2004)


Adjusted R-squared (Wherry (1931))

Bootstrap (Efron (1979))

Cross-validation (Stone (1974), Geisser (1975))


Linear regression

Shibata’s model selector (sms) (1981)

signal-to-noise ratio

test set validation


Akaike information criterion (AIC)

CAT (Parzen, 1974, 1977)

CP (Mallow's Cp, Mallows, 1973)

Deviance information criterion (DIC) (Spiegelhalter, et al., 2002)

FIC (Wei, 1992)

Final prediction error (FPE) (Akaike, 1969)

FPEα (Bhansali and Downham, 1977)

FPEC (De Luna, 1998)

FPER (Larsen and Hansen, 1994)

GM (Geweke and Meese, 1981)

generalized prediction error (GPE) (Moody 1991, Moody 1992)

Hannan and Quinn Criterion (HQ)(Hannan and Quinn, 1979)

KIC, KICc (Cavanaugh, 1999, 2004)

Minimum description length (MDL) (Rissanen, 1978)

Minimum message length (MML) (Wallace and Boulton, 1968)

Predicted squared error (PSE) (Barron, 1984)

PRESS (Allen, 1974)

Schwarz criterion (also Schwarz information criterion (SIC) or Bayesian information criterion (BIC) or Schwarz-Bayesian information criterion) (Schwarz (1978))

Structural risk minimization (SRM) (Vapnik and Chervonenkis (1974))

TIC (Takeuchi's information criterion) (Takeuchi, 1976)

VC-dimension (Vapnik and Chervonenkis (1968, 1971), Vapnik (1979))

Ensemble Methods



Valid XHTML 1.1 | Valid CSS!

Webmaster: Martin Sewell