Recently, I read a paper (you can find it at http://www.research.att.com/~volinsky/bma.html) about bayesian model averaging.
Reading (http://www.stat.colostate.edu/~jah/documents/bma2.ps) the paper, I haven't understood which properties the models need.
I have the impression that if the set of the solutions from all the model isn't convex, it isn't so meaningful to perform this averaging.
If the answers of the different model are very different which is the meaning of this averaging? And if the conceptual uncertainty models are very different? Some people would like to apply this technique to combine different hydrogeological inverted models.......that generally, being an ill-posed problem, have very different probabilistic answer (and very different conceptual models).
But then there is another problem of this method. In order to perform this averaging you need to know the uncertainty of your model.
This, remembering a discussion in GSLIB book (pag 19: uncertainty about uncertainty models), doesn't seem a reasonable task when you have few data (but also from a philosophical point of view!).
Concluding, I have the impression that this method works well with almost exhaustive data set but not when you have few data and the problem is complex (often in geostatistics).
What do you think?
Thank you in advance....
Sebastiano Trevisani Ph.D Padova University Italy
-- * To post a message to the list, send it to [EMAIL PROTECTED] * As a general service to the users, please remember to post a summary of any useful responses to your questions. * To unsubscribe, send an email to [EMAIL PROTECTED] with no subject and "unsubscribe ai-geostats" followed by "end" on the next line in the message body. DO NOT SEND Subscribe/Unsubscribe requests to the list * Support to the list is provided at http://www.ai-geostats.org
