topic Re: Where can I find the G-Optimality map for a custom DoE? in Discussions

Where can I find the G-Optimality map for a custom DoE?

aaidaa — Mon, 09 Oct 2023 14:37:06 GMT

Hi JMP community!

I have conducted a DoE with 4 continuous variables and a 1, 3-level categorical variable with first and second order interactions. The model is a good fit (from ANOVA and lack of fit metrics) but I was hoping to determine areas of greater predictive uncertainty in relation to the output across the design space.

I was therefore wondering how I can find/plot a G-optimality map (i.e. map of prediction uncertainty across the experimental design space) in order to identify the regions of higher uncertainty and run some experiments there.

Thank you in advance!

Re: Where can I find the G-Optimality map for a custom DoE?

Mark_Bailey — Mon, 09 Oct 2023 15:04:02 GMT

As far as I know, the G-optimality is a single value used to evaluate the design as a whole to an ideal design. You can assess the prediction variance after clicking Make Design in the Design Evaluation > Prediction Variance Profiler.

Re: Where can I find the G-Optimality map for a custom DoE?

aaidaa — Fri, 10 Nov 2023 10:43:27 GMT

Sorry just as a follow on from this. I was wondering if there was a way to export the data from the prediction variance profiler ?

Thank you!

Re: Where can I find the G-Optimality map for a custom DoE?

Mark_Bailey — Fri, 10 Nov 2023 14:51:14 GMT

There is no data to export. The plot uses the prediction variance function for a linear model.