Don's Record - Prediction Evaluation
Posted by Petra Challus on December 06, 2002 at 19:03:08:

Hi All,

In one of the strings below we were discussing what Don's prediction record was. After searching through the archives, I did locate the results which are for a 7 month period. Here's the entire post for your review.....Petra





Don's track record for the past 7 months of forecasting
Posted by Lowell on December 08, 2001 at 10:37:20:
Petra has been asking for a summary of Don's predictions - evaluations
So, I gathered together the forecasts he has made to the board and
evaluations (posted and unposted) and the results are below.

I am willing to accept Canie's list of definitions, just so we are
on common ground and because there are no words in these that we
already use to mean something else.

Canie had suggested the following:

1: Surprise? UPE (UnPredictedEarthquke)? or UE? or UF (UnForecasted)

2: MISS - A forecast was made but no corresponding earthquake occurred

3: Obvious - A forecast completely describes a subsequent earthquake and the odds of random success are greater than 1 in 4(?) - 1 in 1-4

4: C Hit - (close hit) A forecast was made and an earthquake occurred which partially matched
the forecast parameters and the odds of occurrence were less than 1 in 5

5: AAA Hit - A forecast completely describes a subsequent earthquake and the odds of random success are less than 1 in 4

in Post:

http://www.earthwaves.org/wwwboard/messages/11721.html

According to these categorizations, here is the summary of Don's
forecasts to this board from May 1 through Dec 6, 2001 according to my
count and evaluation - others may have different numbers, but these are
at least in the ballpark:

Total number of Predictions made: 123

UF (Unforecast) - not applicable
Misses - 2
Obvious - 5
C Hit - 45
AAA Hit - 70

So, we can have confidence that approximately 57% of the time
when Don posts a forecast, an event with random success odds
of less than 1 in 4 will occur within the stated parameters of
the forecast. Likewise his forecast is likely to be off in at
least one parameter 36% of the time. The forecast is likely to
be useless or a complete miss about 5% of the time.


Follow Ups:
     ● Re: Don's Record - Prediction Evaluation - chris in suburbia  15:03:16 - 12/7/2002  (17516)  (0)