The Taguchi method
Article first published online: 15 APR 2011
Copyright © 2011 John Wiley & Sons, Inc.
Wiley Interdisciplinary Reviews: Computational Statistics
Volume 3, Issue 5, pages 472–480, September/October 2011
How to Cite
Mitra, A. (2011), The Taguchi method. WIREs Comp Stat, 3: 472–480. doi: 10.1002/wics.169
- Issue published online: 2 AUG 2011
- Article first published online: 15 APR 2011
- robust design;
- signal-to-noise ratio;
- loss function;
- orthogonal array;
- experimental design
Achieving robustness in product and process designs is of importance to various stakeholders such as manufacturers, suppliers, and consumers. As variability exists in all operations, it is desirable to create products and processes that are not very sensitive to factors that are not controllable. The Taguchi method is an approach to robust design. Inherent in the Taguchi method is the definition of a loss function. This loss function formulation is influenced by the type of quality characteristic under consideration, that is, smaller-is-better, larger-is-better, or target-is-best. Furthermore, based on the selected type of quality characteristic, a performance measure is defined. Such performance measures, usually called signal-to-noise (S/N) ratios, are used to determine optimal settings of the controllable factors. Typically, a two-step procedure is adopted in the Taguchi method. In the first step, the S/N ratio is maximized, whereas in the second step, using an adjustment factor that does not affect the S/N ratio, the mean response is adjusted to meet the target value, where appropriate. Experimental designs make use of orthogonal arrays to determine factor settings for obtaining data for subsequent analysis. The number of experimental runs is very modest in relation to the number of factors being investigated. WIREs Comp Stat 2011 3 472–480 DOI: 10.1002/wics.169
For further resources related to this article, please visit the WIREs website.