Visiting Speaker: Neil Diamond
ACCS Special Seminar
Title: Experimental Design from a Statistical Perspective
Presenter: Neil Diamond
Department of Econometrics and Business
Statistics
Monash University
When: *** MONDAY ***: June 20, 2:00pm
Where: GPS South, Rm: 621 (Access Grid Room)
Seminar Abstract
Statistics can be divided up into two parts: analysis, analysing existing
data, and design, designing how and which data will be collected. Of the
two, design is probably the most important.
The aim of this talk is to give an overview of experimental design from a
statistical viewpoint. In the first part of the talk I will give some of the
history of statistical experimental design which began in agriculture, and
describe how it was adapted by industrial statisticians in the chemical
industry with the development of response surface methodology, and also
discuss further developments that were made by the automotive industry.
In the second part of the talk I will examine the role statistics can play
in the design and analysis of deterministic computer experiments. The
approach that has been most useful is to model the output of the
deterministic computer model as a realization of a stochastic process. A
number of runs of the computer model are made and estimates of the
parameters which describe the stochastic process are made, based on the data
collected. This provides a surrogate model which can be used to estimate the
output of the computer model at untried inputs, and as well giving an
estimate of the error in doing so. A number of examples will be given.
About The Speaker
After graduating with a degree in statistics from Monash University in 1977,
Neil Diamond began work as an industrial statistician at ICI Australia's
Explosive Factory. Later, he was a Senior Research Scientist and Statistics
and Computing Team Leader at the ICI Central Research Laboratories where he
worked on a diverse range of projects. In 1987 he joined the Department of
Mathematics, Computing and Operations Research at Victoria University of
Technology where he taught statistics to undergraduate students, was
industry project director, supervised a number of post-graduate students,
co-ordinated the statistical consulting service as well as consulting to
industry. He spent two six-month stints at the Center for Quality and
Productivity Improvement at the University of Wisconsin-Madison. From 2003
to 2004 he was a Senior Statistician at Insureware Pty Ltd working on the
analysis of long-tailed liabilities, before joining Monash as a Senior
Lecturer and Deputy Director of Consulting in the Department of Econometrics
and Business Statistics. He has a Ph.D. in experimental design from the
University of Melbourne and has published twenty refereed papers on various
aspects of experimental design and statistics.
Title: Experimental Design from a Statistical Perspective
Presenter: Neil Diamond
Department of Econometrics and Business
Statistics
Monash University
When: *** MONDAY ***: June 20, 2:00pm
Where: GPS South, Rm: 621 (Access Grid Room)
Seminar Abstract
Statistics can be divided up into two parts: analysis, analysing existing
data, and design, designing how and which data will be collected. Of the
two, design is probably the most important.
The aim of this talk is to give an overview of experimental design from a
statistical viewpoint. In the first part of the talk I will give some of the
history of statistical experimental design which began in agriculture, and
describe how it was adapted by industrial statisticians in the chemical
industry with the development of response surface methodology, and also
discuss further developments that were made by the automotive industry.
In the second part of the talk I will examine the role statistics can play
in the design and analysis of deterministic computer experiments. The
approach that has been most useful is to model the output of the
deterministic computer model as a realization of a stochastic process. A
number of runs of the computer model are made and estimates of the
parameters which describe the stochastic process are made, based on the data
collected. This provides a surrogate model which can be used to estimate the
output of the computer model at untried inputs, and as well giving an
estimate of the error in doing so. A number of examples will be given.
About The Speaker
After graduating with a degree in statistics from Monash University in 1977,
Neil Diamond began work as an industrial statistician at ICI Australia's
Explosive Factory. Later, he was a Senior Research Scientist and Statistics
and Computing Team Leader at the ICI Central Research Laboratories where he
worked on a diverse range of projects. In 1987 he joined the Department of
Mathematics, Computing and Operations Research at Victoria University of
Technology where he taught statistics to undergraduate students, was
industry project director, supervised a number of post-graduate students,
co-ordinated the statistical consulting service as well as consulting to
industry. He spent two six-month stints at the Center for Quality and
Productivity Improvement at the University of Wisconsin-Madison. From 2003
to 2004 he was a Senior Statistician at Insureware Pty Ltd working on the
analysis of long-tailed liabilities, before joining Monash as a Senior
Lecturer and Deputy Director of Consulting in the Department of Econometrics
and Business Statistics. He has a Ph.D. in experimental design from the
University of Melbourne and has published twenty refereed papers on various
aspects of experimental design and statistics.
