Creating Robust Linear Models Using Generalized Linear Regression (GLR)

• Activity Number: 140
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Quality and Productivity Section
• Author(s): Scott Wise* and Brady Brady
• Companies: JMP (A Division of SAS, Inc.) and JMP (A Division of SAS, Inc.)
• Keywords: Generalized Linear Regression ; Regression ; Stepwise ; Modeling ; Screening

Abstract

Create Robust Linear Models using Generalized Linear Regression (GLR): While not as widely known or used among professionals for creating models on quality and productivity data applications, recent advancement in analytic computing has made GLR models easier to set-up, faster to run, and clearer to analyze.

This means that GLR can now be used to improve upon your traditional linear model method approaches (like Standard Least Squares and Stepwise) in order to create a more robust model, especially for:

- Using GLR to handle complex data, like multicollinearity in your model inputs that traditional methods cannot handle.

- Using GLR to get better screening for the most important factors in a model as compared to traditional methods.

This session will also cover how to best set up GLR modeling to handle a wide range of data types and challenges (binomial, zero-inflated, etc.)