Please use this identifier to cite or link to this item: https://doi.org/10.1080/00031305.2013.817356
Title: On the effect and remedies of shrinkage on classification probability estimation
Authors: Zhang, C.
Liu, Y.
Wu, Z. 
Keywords: Bias
High dimension
Refit
Regularization
Issue Date: 2013
Citation: Zhang, C., Liu, Y., Wu, Z. (2013). On the effect and remedies of shrinkage on classification probability estimation. American Statistician 67 (3) : 134-142. ScholarBank@NUS Repository. https://doi.org/10.1080/00031305.2013.817356
Abstract: Shrinkage methods have been shown to be effective for classification problems. As a form of regularization, shrinkage through penalization helps to avoid overfitting and produces accurate classifiers for prediction, especially when the dimension is relatively high. Despite the benefit of shrinkage on classification accuracy of resulting classifiers, in this article, we demonstrate that shrinkage creates biases on classification probability estimation. In many cases, this bias can be large and consequently yield poor class probability estimation when the sample size is small or moderate. We offer some theoretical insights into the effect of shrinkage and provide remedies for better class probability estimation. Using penalized logistic regression and proximal support vector machines as examples, we demonstrate that our proposed refit method gives similar classification accuracy and remarkable improvements on probability estimation on several simulated and real data examples. © 2013 American Statistical Association.
Source Title: American Statistician
URI: http://scholarbank.nus.edu.sg/handle/10635/105270
ISSN: 00031305
DOI: 10.1080/00031305.2013.817356
Appears in Collections:Staff Publications

Show full item record
Files in This Item:
There are no files associated with this item.

SCOPUSTM   
Citations

6
checked on Oct 7, 2022

WEB OF SCIENCETM
Citations

6
checked on Oct 7, 2022

Page view(s)

121
checked on Oct 6, 2022

Google ScholarTM

Check

Altmetric


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.