How can the report from the PDS response surface calculation be interpreted?
For example, what does VIF mean?
As we perform a regression analysis to create a response surface, there are some assumptions made about the fitted response surface
and the content of this report is used to verify validity of the fitted response surface accordng to the assumptions. VIF is one of the
statistics used to measure how well the fitted response surface complies with the assumptions.
Our response surface is a full quadratic response surface, which includes a constant term, linear terms, pure quadratic terms and cross quadratic
terms. For example, for a two input problem, x1 and x2, the response surface y is a sum of the following terms:
c0 + c1*x1 + c2*x2 + c3*x1*x1 + c4*c2*x2 + c5*x1*x2
For a response surface, regression analysis is an evaluation of the term coefficients, c1, c2, c3, c4 and c5 given some data points.
When evaluating the term coefficients, we want all terms to be uncorrelated with each other. If there exists some correlation, the
estimated term coefficients stend to vary widely from one sample point to another. VIF is a statistical measure for this purpose.
From the report, under the column of VIF, a value of each term between 0.0 and 1.0 is good.