Diagnostics of Generalized Linear Models

• Activity Number: 189
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 10:30 AM to 12:20 PM
• Sponsor: IMS
• Author(s): Heike Schuhmacher*+ and Lutz Duembgen
• Companies: University of Bern and University of Bern
• Keywords: GLM ; model diagnostics ; residual analysis

Abstract

Compared to traditional linear models there is a lack of effective diagnostic methods for generalized linear models. One of the main reasons is that there is no simple counterpart to residual analysis. We propose a new way of transforming the response variables, resulting under a one-point null hypothesis in a standard uniformly distributed sample and not depending on the explanatory variables. The transformed values may be viewed as surrogates for the residuals. The poster gives a detailed description of the method and illustrates its use in various situations.