Department of Industrial Engineering and Management
Ben-Gurion University of the Negev, Israel.
Date :
Wednesday September 12, 2001.
Time :
3:30pm
Address
Burke Science Building
Room:
137
TITLE:
Modeling a Response with Self-Generated and Externally
Generated Sources of Variation - Some Empirical Evidence
ABSTRACT:
A quality practitioner is often required to model the relationship
between a response and the affecting factors where no theory-based models
exist, or are satisfactorily well-developed. In a recent paper, we have
introduced a new approach to model a response, assuming that the latter
experiences both self-generated and externally generated sources of
variation. It was assumed that while both sources transmit to the
response random variation modeled by independent normal variables, the
externally-generated variation includes also a systematic component,
transmitted via a "linear predictor" (a linear combination of the
external affecting factors). The final form of the model is determined by
the data on hand only. In a more recent paper, a maximum likelihood
estimation procedure (with censoring) has been developed for the new
model. In this presentation, we will outline the model, and demonstrate
its versatility in capturing, without resorting to theory-based
arguments, the relationship between a response and the affecting factors.
We do this by some examples from chemical engineering, software
reliability-growth modeling and hardware reliability. The goodness-of-fit
of the resulting models is compared to that of fitted theory-based models
using some published data sets. Implications for various related fields,
like robust parameter design or RSM, are discussed.
About the Speaker
Information about the speaker will be posted here shortly
References
References with background information will be posted here.