Re: Highest Posterior Density


illywhacker wrote:
You can *call* anything a fact, FishBoy, but that changes nothing.

Not in the mind of someone as ignorant as illywhacker.

It makes a difference to those who are much better educated in the
subject matter than illywhacker. They can CHECK the facts
themselves.

It
was rather, or in the alternative, also, invective that had nothing to
do with the substantial issues. All this has arisen because you refuse
to admit that you 1) misinterpreted the OP's question, and 2) given
your misinterpretation, gave poor advice.

The OP was asking a question about finding the maximum of a
posterior distribution in a Bayesian analysis.

Not exactly a deep question.

But apparently one sufficient to derail those who are completely
ignorant about Bayesian inference, such as illywhacker who learned
his mathematics (not even statistics) from physicists, and who
does not even know what a posterior distribution IS and thought
R.A. Fisher was an authority to appeal to, on matters relating to
Riemannian metric and Lebesque measure that are entirely
irrelevant to the Bayesian question, especially at the low level
at which the question was addressed.

RF> I understand the difference between those two perfectly.

Evidently, unlike the rest of the world, you define a probabilty
distribution to be a density function with respect to some underlying
measure. So be it.

I took graduate level courses in Real Analysis and Functional
Analysis which contained plenty of measure theory; in addition
to a graduate course in Measure Theory from the book by
Halmos, all from decent Departments of MATHMATICS, before
I even began studying Statistics.

I did not learn your kind of pseudo mathematics and pseudo
statistics from physicists and computer packages and learned
nothing but some BUZZ WORDS which you don't understand
and thought it had something to do with Bayesian inference.


This still does not change the invariance issue, or
the need to specify the underlying measure.

Those issue never even came up. You were just bluffing on
something you don't understand yourself but heard of those
words somewhere in your physics lab.


RF> illywhacker, give it up.

That was a statement of opinion

Actually, no, it is not an opinion. It is a suggestion or a command.

If was a suggestion based on sound opinion. If it were a
command, I would at least put an exclamation mark after it,
or express it unambiguously as a command.

Which of course I will ignore.

illywhacker;

Which of course will only deepen your state of ignorance about
Bayesian statistics, and statistics in general.

Pity.

-- Reef Fish Bob.

.

Relevant Pages

  • Re: Highest Posterior Density
    ... illywhacker wrote: ... NOT have any basic knowledge about Bayesian methods. ... ignorance about Bayesian statistics and Bayesian inference. ... with 6 posts each by Herman and myself, ...
    (sci.stat.math)
  • Re: prior distributions of estimated parameters
    ... :-) It's ONLY in sci.stat.* can illywhacker, ... COMPLETELY IGNORANT about Bayesian statistics and Bayesian ... RF> in bringing about the renaissance of Bayesian thinking? ...
    (sci.stat.math)
  • Re: prior distributions of estimated parameters
    ... illywhacker wrote much more impertinence without answering questions ... illy> renaissance of Bayesian thinking, ... RF> Have you heard of Lindley, de Finetti, Jeffreys, and L.J. Savage? ... Bayesian statistics ever mention Chernoff, Arrow, or Rubin as ...
    (sci.stat.math)
  • Re: prior distributions of estimated parameters
    ... > COMPLETELY IGNORANT about Bayesian statistics and Bayesian ... > illywhacker, whose debut in USENET was illwhacker, is nothing but ... > a whacko when it comes to STATISTICS. ...
    (sci.stat.math)
  • Re: Statistical ACRONYMS: HOFOSQ, HOSOSQ
    ... to be a Bayesian credible INTERVAL idea, ... ILLY> asked about HPD intervals I believe. ... when Illywhacker stick his nose into my follow-up ... "A generalized Gaussian image model for edge-preserving MAP ...
    (sci.stat.math)