Using SAS/OR's genetic algorithms to optimise PROC NLMIXED's 'hyper parameters'

[csetzkorn]

Hi all,

Just wondering, did someone ever user SAS OR (e.g PROC GA) to optimize the, for example, starting values of PROC NLMIXED? Are there any tutorials white papers out there? Google did not return much ...

Thanks.

Christian

[Rick_SAS]

The "optimal starting value" is the solution to the problem, since that is the starting value that requires zero iterations. 

If you want to read about how to choose good starting guesses solving a maximum likelihood problem with PROC NLMIXED, see

"How to find an initial guess for an optimization" and
"The method of moments: A smart way to choose initial parameters for MLE."