On the relation between the shrinkage effect and a shrinkage method

MaRDI QIDQ1966029zbMATH OpenFDO

Authors Werner Vach

Publication date 2 March 2000

Published in Computational Statistics (Search for Journal in Brave)

zbMATH Keywords

predictionshrinkage effectshrinkage method

Mathematics Subject Classification ID

Generalized linear models (logistic models) (62J12) Inference from stochastic processes and prediction (62M20)

Review text

It is shown that the shrinkage effect is a quite natural phenomenon. It occurs when we use one sample both to construct a predictor and to estimate its future behaviour. That is a reason to use complex techniques like cross validation to reduce the misleading optimism in estimating the future precision of a predictor. The shrinkage methods can be useful in some problems regarding the future behaviour of a predictor. Such a problem is the estimation of the future behaviour of high risk and low risk groups preventing too optimistic expectations about the real difference of their expected outcomes.

Recommendations

• scientific article; zbMATH DE number 3844851
• Publication:4727201
• Shrinkage and Penalized Likelihood as Methods to Improve Predictive Accuracy
• scientific article; zbMATH DE number 4064292
• On reverse shrinkage effects and shrinkage overshoot

Cited in

(3)

• Cross Validation Calibration Based Shrinkage - Revisited for the Construction of Prognostic Indices
• The Shrinkage of Point Scoring Methods
• On reverse shrinkage effects and shrinkage overshoot

This page was built for publication: On the relation between the shrinkage effect and a shrinkage method

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1966029)