A Couple of Questions about convergence using proc model

[bigbigben]

The first question is which convergence measure (R, PPC, RPS, Object and S) the iteration process uses to make decision about whether to stop. My understanding is that usually the criterion should be lower when using object, but it could be higher when the measure is about the change in parameter values.

The second quesiton is when the program will generate a missing value for any of the convergence measures. For example, my program gets a missing RPC, but it is very puzzling, does it mean theta_(k-1) is a missing value?

[PGStats]

Proc Model is part of SAS/ETS. Specialists are more likely to be listening to the SAS Forecasting and Econometrics Community.

[bigbigben]

Hi, thanks for the info. I will copy my post to that forum.