In the article "Beyond the Basics" (April 2012), Govind Ramu wrote, "There is very little reference material on matrix data analysis itself." On the contrary, there are many books on linear modeling and years of articles in statistical journals that establish these techniques.1
The article refers to step-by-step approaches, which include stepwise linear regression2 and stepwise logistic regression. For building trees, you should consider the classification and regression tree, a data-mining tree-building method.
For risk management, you should consider using maximum likelihood estimation of lifetime distribution parameters for multiply censored lifetime data.3
You also could include various covariates to account for accelerated lifetime modeling. And Bayesian methods are becoming more popular in decision making.4
My opinion is that ASQ journals such as the Journal of Quality Technology and Technometrics include many useful techniques that are not mentioned frequently enough in the quality tools discussion.
References and Notes
1. Some books to consider include Statistics for Experimenters: Design, Innovation and Discovery, second edition, by George E.P. Box, J. Stuart Hunter and William G. Hunter (Wiley-Interscience, 2005); Empirical Model-Building and Response Surfaces by George E.P. Box and Norman R. Draper (Wiley, 1987); Response Surface Methodology by Raymond H. Myers and Douglas C. Montgomery (Wiley-Interscience, 1995); and Variance Components by Shayle R. Searle, George Casella and Charles E. McCulloch (Wiley-Interscience, 2006).
2. For more, see Applied Linear Statistical Models by John Neter, Michael Kutner, William Wasserman and Christopher Nachtsheim (McGraw-Hill, 1996).
3. For guidance in this area, see Statistical Methods Reliability Data by William Meeker and Luis A. Escobar (Wiley-Interscience, 1998); Applied Reliability, third edition, by Paul A. Tobias and David Trindade (Chapman and Hall, 2011); and JMP 10 Quality and Reliability Methods from the SAS Institute (SAS Publishing, 2012).
4. For more, see Statistical Decision Theory and Bayesian Analysis by James O. Berger (Springer, 2010); Data Analysis Using Regression and Multilevel/Hierarchical Models by Andrew Gelman and Jennifer Hill (Cambridge University Press, 2006); and Bayesian Data Analysis by Andrew Gelman, John B. Carlin, Hal S. Stern and Donald B. Rubin (Chapman and Hall, 2003).
Thanks for the detailed feedback. The "very little reference" I mentioned in my article refers to finding answers for the questions, "What is matrix data analysis?" and "What did the Union of Japanese Scientists and Engineers (JUSE) originally consider as matrix data analysis?"
I did extensive bibliographical research using the ASQ Knowledge Center historical archives, a quality information search, a Google search, Wikipedia and my own collection of more than 100 quality books.
The only resource I found that refers to matrix data analysis is Management for Quality Improvement.1 It’s also the only one that explains it as a collection of multivariate analysis tools. In the book, Shigeru Mizuno used principal components analysis as an example to explain matrix data analysis.
After identifying the tools JUSE considered as matrix data analysis, there are plenty of reference materials available for the individual tools. But QP is a magazine that covers a wide range of quality topics from various industry sectors, and it’s targeted at readers of all levels of experience and education.
San Jose, CA
1. Shigeru Mizuno, Management for Quality Improvement, Productivity Press, 1988.