Subscribe Bookmark RSS Feed

Prior distribution in nonlinear DOE


Community Member


Jan 28, 2017


I'm trying to make a design for a nonlinear model.
Some prior knowledge of the parameters is available: a guess for the mean of the parameters as well as a guess for the variance-covariance matrix.
I've used the mean to initialise the model, but I'm not sure on how to use the variance-covariance matrix.
On the following picture you can see the option to fill in values for the parameters:

  • What happens if I fill in 95% confidence intervals for all my parameters?
  • If I put number of radii to zero in advanced options (local optimal design) will these parameter values still somehow be used?
  • I have strong correlations between parameters in my prior distribution. Is there a way JMP can (or does) take this into account?