GENERALIZED LINEAR-MODELS AND PROCESS VARIATION

Several approaches have been proposed for optimizing both the mean and variation of a process simultaneously. This paper reviews some of the se methods and studies ways in which generalized linear models can be adapted for use with them. Specifically, a generalized linear model wi th gamma error distribution and log link function is used to model var iation as (1) part of a screening method for variance control factors and (2) part of an algorithm for simultaneous maximum likelihood estim ation of mean and variance parameters. The advantages and disadvantage s of these two approaches are examined in detail and compared to other current methods.