Personen | Personen nieuwe site | Google | Route | Contact | AfdrukkenLogin 
Design Of Experiments  
Public Defence Bradley Jones
The first day of the conference is concluded with the public defence of the doctoral thesis of Bradley Jones. All conference participants are warmly invited to attend the defence and the reception afterwards. The abstract for the defence is given below.
Computer Aided Designs for Practical Experimentation
The first uses of designed experiments in agriculture required a statistician to create a design and analyze the data. The statistician was not a scarce resource because agricultural experiments require a year to run so there was plenty of time for statisticians to plan all studies for an entire experimental station. As DOE has transcended its roots in agriculture, the scarcity of statisticians relative to the number of potential applications of DOE made professional statistical involvement in every experimental study infeasible. One way of dealing with this scarcity is to use tabled sample plans that are available in textbooks on designed experiments. These plans typically have very desirable statistical properties and have a fairly wide range of applicability. This "one size fits all" approach to design choice is quite popular. Many short courses in DOE teach this approach exclusively. In the last two decades, however, a viable alternative has arisen due to huge increase in the power of computers and their accompanying software. A computer-aided approach to DOE offers the possibility of a tailored approach to design as opposed to the "one size fits all" approach. This talk demonstrates the utility of the computer-aided tailored approach to DOE using a real example from Belgian automobile manufacturing. Unique aspects of the problem made the use of any previously tabled design impossible. The talk also addresses some technical concerns about the use of optimization algorithms to produce designs for general engineering and scientific use.