On Wed, 28 Apr 2004, Eddie Jaye wrote in part:
> I am looking for a solution to a Bayesian algorithm problem that
> concerns calculating additional weighting inputs. Bayes' theorem tells
> me ... <snip, discourse...>
> ... each of these individual conditions may have their own weighting
> value, related to their importance. How then do I include this
> weighting as part of the prior value input for the next probability? I
> have tried to cut and paste to give a diagrammatic representation of
> what I mean, but it did not come out properly
<snip, abortive diagram>
> Above ss1, ss2 and ss3 are independent of each other but each has and
> additional weighting value in terms of importance.
The diagram looked as though it were trying to represent a dendrogram,
showing various branches of a decision tree (or equivalent). If that's
what was intended, {ss1, ss2, ss3} are not independent, they're disjoint
(i.e., mutually exclusive). And you do not indicate the events whose
probabilities are supposed to depend on the various branches of the
tree.
When you speak of "weighting", do you mean a measure of importance that
is assigned arbitrarily and a priori, or a measure that is to be
inferred from observing data, or something else?
And when you write "How then do I include this weighting as part of the
prior value input for the next probability?", are you asking how to
calculate weights from data, or how to represent weights on the
dendrogram? If what you want to end up with is an algorithm, it might
be helpful if you told us how far you've gotten with devising an
algorithm so far.
-- DFB.
------------------------------------------------------------
Donald F. Burrill [EMAIL PROTECTED]
56 Sebbins Pond Drive, Bedford, NH 03110 (603) 626-0816
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
