RE: Models that abort before convergence Addendum

From: Mark Sale - Next Level Solutions <mark>
Date: Fri, 21 Nov 2008 12:09:58 -0700
 Leonid,  Let me understand:You now have a theor= y that the way to determine whether the NONMEM error messages are useful (i= .e., they tell you something about the model "goodness") is a poll.  T= his, I think is a theory (and one well established in epistomolgy) of how t= o find an optimal solution - appeal to a large number of presumably well in= formed people.  As data that may be relavant to this theory, I would p= oint out that a poll gave us GW Bush as our 43rd president. Nick, = in contrast has suggested that the error messages could be used as a source= of random numbers.   This also, I think, is a theory without dat= a to support or contradict it.   So .... Let me propose a= solution - let's generate some data.  Suppose we randomly generate 10= 00 models.  We could tests the hypotheses:Are the error messag= es random (I suspect they are not, that there is some information in them).=   To test this, see if the error messages are predictive of other (pre= sumably non-random) measure of goodness - NPC and NPDE, and perhaps PPC com= e to mind.Do the error messages provide information not readily ava= ilable in NPC, NPDE and PPC.Not really sure how to test this, without s= ome "gold standard" of goodness, except perhaps to compare the different me= asures to the model that was used to simulate the data (seems like measures= based on that would be "correct" in some way??).  I need some ideas o= n this.I can generate, run and extract results from random mode= ls (using the GA software) - I already have NPDE and PPC in it, was thinkin= g of adding NPC.Any interest/collaborators??Mar= k Sale MD Next Level Solutions, LLC www.NextLevelSolns.com 919-846-9185 -------- Original Message -------- Subject: Re: [NMusers] Models that abort before convergence Addendum From: Leonid Gibiansky