Data Analysis Papers
Applications of the GLMSELECT Procedure for Megamodel Selection
Cohen, Robert; SAS Institute 2009
Abstract
When you can select regression models from tens of thousands of effects, what possibilities for modeling are open to you? This paper explores applications of the GLMSELECT procedure in SAS/STAT® software to such problems. The GLMSELECT procedure supports a variety of model selection methods for general linear models. Examples of megamodels arising in genomic data analysis and nonparametric modeling are discussed. In addressing these examples, built-in facilities of the procedure to handle validation and test data are highlighted in addition to techniques for extending the procedure’s functionality to address model selection bias by using bootstrap-based model averaging.
Statistics and Operations
Research Home Page