Partial Factor Modeling: Predictor-Dependent Shrinkage for Linear Regression

• 成果类型: Article
• 署名作者: Hahn, P. Richard; Carvalho, Carlos M.; Mukherjee, Sayan
• 署名单位: University of Chicago; University of Texas System; University of Texas Austin; Duke University; Duke University
• 刊物名称: JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
• ISSN/ISSBN: 0162-1459
• DOI: 10.1080/01621459.2013.779843
• 发表日期: 2013
• 页码: 999-1008
• 关键词: principal component; variable selection; gene-expression; arbitrage; distributions

摘要

We develop a modified Gaussian factor model for the purpose of inducing predictor-dependent shrinkage for linear regression. The new model predicts well across a wide range of covariance structures, on real and simulated data. Furthermore, the new model facilitates variable selection in the case of correlated predictor variables, which often stymies other methods.