Re: Petra's predictions

On Jan 19, 3:29 am, Petra <petras...@xxxxxxxxxxx> wrote:
On Jan 18, 9:36 pm, "Mike Williams" <miklw...@xxxxxxxxxxx> wrote:



<rog...@xxxxxxxxxxxxxxx> wrote in message

news:808b8319-fc1b-4075-8efd-651324bc1219@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Hi all;

Starting a new thread for this.

Petra has been making some unpleasant remarks about me for no good
reason. I've let it slide because I know she doesn't understand the
analytical process too well.

But her latest post gives some misleading numbers and I'd like to know
how she comes up with them. A claim of 95%-99% better than chance
requires some support as it isn't derived from the hit/miss numbers
she also posted.

Petra?

Roger

My interpretation was that the 95%-99% figure was most likely the confidence
level - i.e. that somebody initially might have told her that her
predictions were better than chance, at that level of confidence.

Mike Williams
Arroyo Grande, CA USA- Hide quoted text -

- Show quoted text -

Hi Mike,

The figure comes from an expert and the final report as I've said
before will be published this Spring. So Ya'll are just gonna have to
wait and read it then.

Whatever is said above no doubt will change tomorrow, next week, a
month from now and maybe next year. Hunter has only stood firm on one
number in the ten years I've known him, so we should anticipate
further changes and most likely after the report is published.

Stay tuned for April/May 2008. LOL

Petra

Petra;

One thing I've never gotten you to undestand is that the output
depends on the input.

For example, I got the above results using the 1973-2006 time period
to get probabilities. I reran it using the 2000-2006 time period and
found that the unchanged predictions got a rating of -2.118 (way below
chance) whereas the elimination of small mags, aftershocks and
overlaps raised the answer to -0.380, still below chance but only a
little.

Now which answer is correct?

All of them. They're all correct for the input that was used to get
them.

The question then becomes "Is the input suitable for the problem?" and
that of course can be debated.

The long database gives a long term answer to probability. The shorter
run adjusts for recent conditions. An argument can be made for using
the current year for each prediction which best compensates for large
changes in seismicity.

Roger
.

Relevant Pages

  • Re: Science of choices falls out of research
    ... in making certain predictions. ... That probabilities are only in the mind is merely philosophical add on. ... not a situation where I have a chance of going up and a chance of going ... the probability that the top card is the ace of spades, ...
    (talk.origins)
  • Re: Science of choices falls out of research
    ... Which merely states that statistics approximate probabilities ... one supposes that the chance is solely in the mind of the observer. ... Before revealing what is behind that curtain, the host reveals a goat ...
    (talk.origins)
  • Re: Numbers to think about
    ... real positive group or the false positive group are not equal. ... there is a 99% chance that it's a real ... the probabilities of Floyd's results being a false positive change ...
    (rec.bicycles.racing)
  • How I explain supernatural reality
    ... I decided to use the mathematical tool of chance and probabilities. ... I apply this argument to the prediction of future historical events. ... The likelihood of these correct predictions occurring by chance is ... I believe that the designer is a prescient intelligence possibly from ...
    (alt.religion.christian)
  • How I explain supernatural reality
    ... I decided to use the mathematical tool of chance and probabilities. ... I apply this argument to the prediction of future historical events. ... The likelihood of these correct predictions occurring by chance is ... I believe that the designer is a prescient intelligence possibly from ...
    (alt.religion.christian.roman-catholic)