Efforts to improve quality require that the factors affecting quality be identified. When the common assumptions associated with the normal scenario are not met two alternative approaches are commonly pursued: Normalization of data and the use of generalized linear models. In this paper we introduce a third approach that attempts to model the unknown distribution of the underlying quality variable via Inverse Normalizing Transformations (INTs), recently developed. Since the new approach uses a uniform procedure (irrespective of the underlying distribution) no specification is needed of either the required normalizing transformation (the first approach) or of the error distribution and the link function (the second approach). An application of the new methodology to published experimental data from the Semiconductor industry is demonstrated.
|Title of host publication||International Conference on Modeling and Analysis of Semiconductor Manufacturing (MASM 2000)|
|Editors||Jeffery K. Cochran, John W. Fowler, Steven Brown|
|State||Published - 2000|