"Hatunen" <hatu...@cox.net> wrote in message

news:cdnkp3d7b7jcmvcd9dvauag2h6rqe0b1uk@4ax.com...



On Fri, 25 Jan 2008 11:07:07 -0800 (PST), rog...@lpbroadband.net
wrote:

Hi all;

I'm looking for input on this.

A number of predictors use bullseye predictive schemes, notably JOB
and Petra. This involves partial credit for quakes which fall outside
the primary prediction parameters by specified amounts.

I've tried several ways of dealing with this. I have insisted on
perfect hits with Petra and I've used multiple probabilities with JOB
but neither method satisfies anyone. It's difficult to deal with
because accurate probability is essential to determining significance
but dealing with 4 ranges with 3 parameters is a LOT of possible
answers.

I'm thinking that the proper way to handle this might be to consider
such a prediction to be multiple predictions, one for each possible
combination, since the predictor isn't specifying which combination to
expect. That would be devastating to their significance, as even a
perfect hit would also involve multiple misses as well. But it's
mathematically sound as far as I can tell.

The best solution is to make only specific predictions and accept only
perfect hits but that's not something I can control. It's up to the
predictors to quit wanting credit for "near enough" results.
Especially when "near enough" is quite a way away.

Any opinions?

Yeah. They cover this stuff in university statistics courses. You
ought to try taking some of them.

Unfortunately, statistics courses don't deal with logarithmic
data like earthquake M or Richter sizes so calculating standard
deviations won't work real well. Of course you could change them
to non-logartihmic numbers. That might give a better idea of what
you're actually facing.

--
************* DAVE HATUNEN (hatu...@cox.net) *************
* Tucson Arizona, out where the cacti grow *
* My typos & mispellings are intentional copyright traps *

Dave, I think Roger is well-qualified to be doing the statistical analysis.
If I'm not mistaken, he spent a career doing something like that for the
USGS.

Roger? Could you, once again, re-state your relevant background for us?

Mike Williams
Arroyo Grande, CA U.S.

On Fri, 25 Jan 2008 11:07:07 -0800 (PST), rog...@lpbroadband.net
wrote:



Hi all;

I'm looking for input on this.

A number of predictors use bullseye predictive schemes, notably JOB
and Petra. This involves partial credit for quakes which fall outside
the primary prediction parameters by specified amounts.

I've tried several ways of dealing with this. I have insisted on
perfect hits with Petra and I've used multiple probabilities with JOB
but neither method satisfies anyone. It's difficult to deal with
because accurate probability is essential to determining significance
but dealing with 4 ranges with 3 parameters is a LOT of possible
answers.

I'm thinking that the proper way to handle this might be to consider
such a prediction to be multiple predictions, one for each possible
combination, since the predictor isn't specifying which combination to
expect. That would be devastating to their significance, as even a
perfect hit would also involve multiple misses as well. But it's
mathematically sound as far as I can tell.

The best solution is to make only specific predictions and accept only
perfect hits but that's not something I can control. It's up to the
predictors to quit wanting credit for "near enough" results.
Especially when "near enough" is quite a way away.

Any opinions?

Yeah. They cover this stuff in university statistics courses. You
ought to try taking some of them.

Unfortunately, statistics courses don't deal with logarithmic
data like earthquake M or Richter sizes so calculating standard
deviations won't work real well. Of course you could change them
to non-logartihmic numbers. That might give a better idea of what
you're actually facing.

--
************* DAVE HATUNEN (hatu...@cox.net) *************
* Tucson Arizona, out where the cacti grow *
* My typos & mispellings are intentional copyright traps *

Hi all;

I'm looking for input on this.

A number of predictors use bullseye predictive schemes, notably JOB
and Petra. This involves partial credit for quakes which fall outside
the primary prediction parameters by specified amounts.

I've tried several ways of dealing with this. I have insisted on
perfect hits with Petra and I've used multiple probabilities with JOB
but neither method satisfies anyone. It's difficult to deal with
because accurate probability is essential to determining significance
but dealing with 4 ranges with 3 parameters is a LOT of possible
answers.

I'm thinking that the proper way to handle this might be to consider
such a prediction to be multiple predictions, one for each possible
combination, since the predictor isn't specifying which combination to
expect. That would be devastating to their significance, as even a
perfect hit would also involve multiple misses as well. But it's
mathematically sound as far as I can tell.

The best solution is to make only specific predictions and accept only
perfect hits but that's not something I can control. It's up to the
predictors to quit wanting credit for "near enough" results.
Especially when "near enough" is quite a way away.

Any opinions?

Roger

On Fri, 25 Jan 2008 17:36:28 -0800 (PST), rog...@lpbroadband.net
wrote:



On Jan 25, 4:59 pm, "Mike Williams" <miklw...@pacbell.net> wrote:
"Hatunen" <hatu...@cox.net> wrote in message
Yeah. They cover this stuff in university statistics courses. You
ought to try taking some of them.

Unfortunately, statistics courses don't deal with logarithmic
data like earthquake M or Richter sizes so calculating standard
deviations won't work real well. Of course you could change them
to non-logartihmic numbers. That might give a better idea of what
you're actually facing.

Dave, I think Roger is well-qualified to be doing the statistical analysis.
If I'm not mistaken, he spent a career doing something like that for the
USGS.

Roger? Could you, once again, re-state your relevant background for us?

Mike;

Academically, a BA in Psychologyy but I had started out in Geology so
had enough credits to qualify as a geophysicist with the government.

I spent 34 years, most of it with the Seismology group in first NOAA
and then USGS when they took us over. As part of that I created and
ran the Prediction Monitoring and Evaluation Program which accepted
predictions from all sources and evaluated their success rates. The
project was terminated prematurely so I never got to publish a final
report but after looking at several hundred predictors, none managed
to pass the test.

My current programs are similar to the PMEP program but use better
statistical methods adopted from Dr Alan Jones. I was also able to
persuade him to create the method to be used on predictions with
different probabilities.

Then why are you asking these questions here in this group?
You're obviously the expert.

--
************* DAVE HATUNEN (hatu...@cox.net) *************
* Tucson Arizona, out where the cacti grow *
* My typos & mispellings are intentional copyright traps *

Hi all;

I'm looking for input on this.

A number of predictors use bullseye predictive schemes, notably JOB
and Petra. This involves partial credit for quakes which fall outside
the primary prediction parameters by specified amounts.

I've tried several ways of dealing with this. I have insisted on
perfect hits with Petra and I've used multiple probabilities with JOB
but neither method satisfies anyone. It's difficult to deal with
because accurate probability is essential to determining significance
but dealing with 4 ranges with 3 parameters is a LOT of possible
answers.

I'm thinking that the proper way to handle this might be to consider
such a prediction to be multiple predictions, one for each possible
combination, since the predictor isn't specifying which combination to
expect. That would be devastating to their significance, as even a
perfect hit would also involve multiple misses as well. But it's
mathematically sound as far as I can tell.

The best solution is to make only specific predictions and accept only
perfect hits but that's not something I can control. It's up to the
predictors to quit wanting credit for "near enough" results.
Especially when "near enough" is quite a way away.

Any opinions?

On Fri, 25 Jan 2008 17:39:25 -0800 (PST), rog...@lpbroadband.net
wrote:

Unfortunately, statistics courses don't deal with logarithmic
data like earthquake M or Richter sizes so calculating standard
deviations won't work real well. Of course you could change them
to non-logartihmic numbers. That might give a better idea of what
you're actually facing.

Doesn't matter. The logarithmic nature of magnitude has no bearing on
the analysis which is concerned only with hit or miss results.

It matters if a hit can be an order of maghnitude outside the
main predicted number. To say that a quake of magnitude 7 will
occur and then allowing a quake of magnitude 6 to fulfill the
prophecy is to be deceived by the "small" difference of "only"
one.

But yes, if you set up very clear hit or miss criteria and don't
fudge them post facto it doesn't much matter. Nevertheless,
predicting a quake between, say, 4.0 and 7.0 is kind of silly.

--
************* DAVE HATUNEN (hatu...@cox.net) *************
* Tucson Arizona, out where the cacti grow *
* My typos & mispellings are intentional copyright traps *

On Fri, 25 Jan 2008 11:07:07 -0800 (PST), rogerh@lpbroadband.net
wrote:

Hi all;

I'm looking for input on this.

A number of predictors use bullseye predictive schemes, notably JOB
and Petra. This involves partial credit for quakes which fall outside
the primary prediction parameters by specified amounts.

I've tried several ways of dealing with this. I have insisted on
perfect hits with Petra and I've used multiple probabilities with JOB
but neither method satisfies anyone. It's difficult to deal with
because accurate probability is essential to determining significance
but dealing with 4 ranges with 3 parameters is a LOT of possible
answers.

I'm thinking that the proper way to handle this might be to consider
such a prediction to be multiple predictions, one for each possible
combination, since the predictor isn't specifying which combination to
expect. That would be devastating to their significance, as even a
perfect hit would also involve multiple misses as well. But it's
mathematically sound as far as I can tell.

The best solution is to make only specific predictions and accept only
perfect hits but that's not something I can control. It's up to the
predictors to quit wanting credit for "near enough" results.
Especially when "near enough" is quite a way away.

Any opinions?

Yeah. They cover this stuff in university statistics courses. You
ought to try taking some of them.

Unfortunately, statistics courses don't deal with logarithmic
data like earthquake M or Richter sizes so calculating standard
deviations won't work real well. Of course you could change them
to non-logartihmic numbers. That might give a better idea of what
you're actually facing.

--
************* DAVE HATUNEN (hatunen@cox.net) *************
* Tucson Arizona, out where the cacti grow *
* My typos & mispellings are intentional copyright traps *

On Jan 25, 4:59 pm, "Mike Williams" <miklw...@pacbell.net> wrote:
"Hatunen" <hatu...@cox.net> wrote in message

Yeah. They cover this stuff in university statistics courses. You
ought to try taking some of them.

Unfortunately, statistics courses don't deal with logarithmic
data like earthquake M or Richter sizes so calculating standard
deviations won't work real well. Of course you could change them
to non-logartihmic numbers. That might give a better idea of what
you're actually facing.

Dave, I think Roger is well-qualified to be doing the statistical analysis.
If I'm not mistaken, he spent a career doing something like that for the
USGS.

Roger? Could you, once again, re-state your relevant background for us?

Mike;

Academically, a BA in Psychologyy but I had started out in Geology so
had enough credits to qualify as a geophysicist with the government.

I spent 34 years, most of it with the Seismology group in first NOAA
and then USGS when they took us over. As part of that I created and
ran the Prediction Monitoring and Evaluation Program which accepted
predictions from all sources and evaluated their success rates. The
project was terminated prematurely so I never got to publish a final
report but after looking at several hundred predictors, none managed
to pass the test.

My current programs are similar to the PMEP program but use better
statistical methods adopted from Dr Alan Jones. I was also able to
persuade him to create the method to be used on predictions with
different probabilities.

Unfortunately, statistics courses don't deal with logarithmic
data like earthquake M or Richter sizes so calculating standard
deviations won't work real well. Of course you could change them
to non-logartihmic numbers. That might give a better idea of what
you're actually facing.

Doesn't matter. The logarithmic nature of magnitude has no bearing on
the analysis which is concerned only with hit or miss results.

On Fri, 25 Jan 2008 17:36:28 -0800 (PST), rogerh@lpbroadband.net
wrote:

On Jan 25, 4:59 pm, "Mike Williams" <miklw...@pacbell.net> wrote:
"Hatunen" <hatu...@cox.net> wrote in message

Yeah. They cover this stuff in university statistics courses. You
ought to try taking some of them.

Unfortunately, statistics courses don't deal with logarithmic
data like earthquake M or Richter sizes so calculating standard
deviations won't work real well. Of course you could change them
to non-logartihmic numbers. That might give a better idea of what
you're actually facing.

Dave, I think Roger is well-qualified to be doing the statistical
analysis.
If I'm not mistaken, he spent a career doing something like that for the
USGS.

Roger? Could you, once again, re-state your relevant background for us?

Mike;

Academically, a BA in Psychologyy but I had started out in Geology so
had enough credits to qualify as a geophysicist with the government.

I spent 34 years, most of it with the Seismology group in first NOAA
and then USGS when they took us over. As part of that I created and
ran the Prediction Monitoring and Evaluation Program which accepted
predictions from all sources and evaluated their success rates. The
project was terminated prematurely so I never got to publish a final
report but after looking at several hundred predictors, none managed
to pass the test.

My current programs are similar to the PMEP program but use better
statistical methods adopted from Dr Alan Jones. I was also able to
persuade him to create the method to be used on predictions with
different probabilities.

Then why are you asking these questions here in this group?
You're obviously the expert.

--
************* DAVE HATUNEN (hatunen@cox.net) *************
* Tucson Arizona, out where the cacti grow *
* My typos & mispellings are intentional copyright traps *

data like earthquake M or Richter sizes so calculating standard
deviations won't work real well.

Then why are you asking these questions here in this group?
You're obviously the expert.

It matters if a hit can be an order of maghnitude outside the
main predicted number. To say that a quake of magnitude 7 will
occur and then allowing a quake of magnitude 6 to fulfill the
prophecy is to be deceived by the "small" difference of "only"
one.

But yes, if you set up very clear hit or miss criteria and don't
fudge them post facto it doesn't much matter. Nevertheless,
predicting a quake between, say, 4.0 and 7.0 is kind of silly.

Hatunen <hatunen@cox.net> wrote in news:d35lp3143qds67mb7lt9t6cjglalondflv@
4ax.com:

Then why are you asking these questions here in this group?
You're obviously the expert.

Do experts have all the answers? No they don't. You need to check
your definition of "expert".

On Sat, 26 Jan 2008 04:26:09 -0000, Skywise
into@oblivion.nothing.com> wrote:

Hatunen <hatunen@cox.net> wrote in news:d35lp3143qds67mb7lt9t6cjglalondflv@
4ax.com:

Then why are you asking these questions here in this group?
You're obviously the expert.

Do experts have all the answers? No they don't. You need to check
your definition of "expert".

Here???

It is to laugh...