Hotelling's T2: A Multivariate Statistic for Industrial Process Control


Mason, Robert L.; Young, John C.   (1998, ASQ)   Southwest Research Institute, San Antonio, TX; McNeese State University, Lake Charles, LA

Annual Quality Congress, Philadelphia, PA    Vol. 52    No. 0
QICID: 10666    May 1998    pp. 78-85
List $10.00
Member $5.00

You will need to be signed in.
New to ASQ? Register here.

Article Abstract

Statistical process control in the multivariate environment is served by the flexibility and versatility of Hotelling's T2 statistic. Valuable characteristics of this tool include the ease with which it can be charted and interpreted as well as its applicability to time dependent and independent observations, to small and consistent process changes, and to real time processes. T2 is expressed in terms of parameter estimates of an underlying normal distribution in p-dimensional space, based on an historical data set. Signal interpretation can be handled by the MYT (Mason, Young, Tracy) decomposition, which generates both unconditional and conditional terms. The unique conditional terms of this decomposition enhance the statistic's detection of small, consistent process shifts. A limitation of T2 is that it will not necessarily make a distinction between variability shifts and mean shifts of the variable(s) that caused the signal.


Multivariate quality control,Statistical process control (SPC),Process control,Process analysis

Browse QIC Articles Chronologically:     Previous Article     Next Article

New Search

Featured advertisers

ASQ is a global community of people passionate about quality, who use the tools, their ideas and expertise to make our world work better. ASQ: The Global Voice of Quality.