Taguchi methods (created 2005-02-22).

This page is moving to a new website.

Genichi Taguchi was a Japanese engineer and statistician who developed a wide range of statistical tools for improving the quality of industrial manufacturing. These tools are collectively known as Taguchi methods. One of the most powerful of these tools is robust product design. In this design, certain manufacturing parameters are varied in an inner array, and environmental parameters are varied in an outer array. The goal is to find a combination of manufacturing parameters that perform well (on target with minimal variation) across all of the environmental parameters. Such a combination is robust across the range of conditions that the product might be used at.

An excellent summary appears in section 5.5.6 of the Engineering Statistics Handbook published by the U.S. National Institute of Standards and Technology.

Update (May 3, 2005): Here are some references on Taguchi designs that were posted on edstat-l.