文档介绍：Six Sigma PLAYBOOKPLAYBOOKPLAYBOOK
Topic 23
Correlation and Regression viding an indication of how strong the rela-
Analysis tionship is between them.
How
Overview 1. Define the variables, X and Y. There
should be some connection or association
Why between them. It may be time or some
Can you effectively predict future performance physical entity (mon machine or op-
using past results? How much confidence can erator), but there should be an association to
you place in your predictions? Correlation and assure that there is some relevance to the
Regression are two tools, used in conjunction prediction.
with scatter plot diagrams to answer these ques-
2. Draw a scatter plot of X and Y. The plot
tions. As you might guess, the answer depends
should show a relationship between X and
upon the process and the data.
Y. Generally, the relationship should be
Correlation is used to measure the association linear. It may not be perfect; the regres-
between two or more variables. If you wish to sion analysis problem is to find the best
determine which input variables are key factors straight line prediction. This is our
that result in producing a high quality product mathematical model. Look at the illustra-
(the Vital Few), correlation is a way to take a tions on the following page to see some of
quick assessment of their relationship. the types