On 3/19/19 12:49 PM, Jeff Newmiller wrote:
Highly off topic. Try StackOverflow.
As it stands it's off-topic for SO. (You would just be making more work
for those of us who know the rules but need 4 close votes for
migration.) Better would be immediately posting at CrossValidated.com
(i.e., stats.stackexchange.com)
--
David.
On March 19, 2019 10:42:24 AM PDT, Philip Rhoades <p...@pricom.com.au> wrote:
People,
I have only a general statistics understanding and have never actually
used Bayes' Theorem for any real-world problem. My interest lies in
developing some statistical approach for addressing the subject above
and it seems to me that BT is what I should be looking at? However,
what I am specifically interested in is how such a work-up would be
developed for a year-on-year situation eg:
I think it is likely that TEHTC could be triggered by a multi-gigaton
release of methane from the Arctic Ocean and the Siberian Permafrost in
any Northern Hemisphere Summer from now on (multiple physical and
non-physical, human positive feedback loops would then kick in).
So, say my estimate (Bayesian Prior) is that for this coming (2019) NHS
the chance of this triggering NOT occurring is x%. The manipulation is
then done to calculate the posterior for 2019 - but for every
successive
year (given the state of the world), isn't it true that the chance of a
triggering NOT occurring in the NHS MUST go down? - ie it is just an
argument about the scale of the change from year to year?
It seems to be that the posterior for one year becomes the prior for
the
next year? Once the prior gets small enough people won't bother with
the calculations anyway . .
Does anyone know of any existing work on this topic? I want to write a
plain-English doc about it but I want to have the stats clear in my
head
. .
Thanks,
Phil.
______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
