How to get BayesOpt to reproduce each Batch?

Question

Hi All,

Just trying to prep some BayesOpt demos - how do you get the BayesOpt to recreate the model that's formed with each Batch? For example I will run my initial batches, get a model with a RSq and suggested conditions for the next batch and go ahead and run it. If I then re-run the BayesOpt on that batch again (say after I've assembled a completed BayesOpt run and hide-exclude the rows) I end up with different Rsq and suggested runs - any suggestions?

Thanks,
Ben

MikeD_Anderson · Answer

You can also try putting

..., Set Random Seed( 1234 ), ...

Into the saved script for BayesOpt itself.  When a platform uses a random seed that's often a scriptable option even if it's not documented in the scripting index.

Phil_Kay · Answer

I know our summer (now year-round!) intern @bradley_bowen struggled with this stuff. I'm sure he could help.

Ben_BarrIngh · Answer

Managed to figure it out! The BayesOpt is being made from a seed that is set as a random number (which will point to what candidate sets are chosen, what next steps it suggests etc.) - so you can add a table script for 'set seed' which is 'Random Reset ([your number])' - just make sure you click it before running BO and it will be the same!