Finding the Best Model: To Use GLMSELECT, GLM, or REG ... That Is the Question!

• Activity Number: 413
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section for Statistical Programmers and Analysts
• Author(s): Heather Murphy*+
• Companies: Eli Lilly and Company
• Keywords: SAS ; GLMSELECT ; Model Selection

Abstract

The GLMSELECT procedure is new in SAS 9.2. GLMSELECT is a flexible and innovative statistical programming procedure that is used for model selection in the construct of general linear models. GLMSELECT incorporates favorable attributes from GLM and REG procedures while offering a wide assortment of modern model selection methods. This new addition allows the statistician to include numerous effects (tens of thousands) in the model and is capable of providing both numerical and graphical results. I will share practical suggestions of how to use GLMSELECT to choose a preliminary model, and then explore the results in more detail. Examples will be provided using this new modeling method, including a variety of innovative select, stop, and choose options. GLMSELECT is a procedure that can be beneficial to all statisticians, we just need to know how to use it... and that's where I come in.