J C-W wrote in message
news

"Franz Heymann" wrote in message
...
The following is true (I'm a sad soul). At the beginning of the year I
did
a
small experiment noting forecasts and noting what actually happened
for
my
city over 3 weeks (I got bored after 3 weeks). I collected each
forecast
for
the following day only, no long range stuff. The three sites were
Yahooo,
UK Weather.com and the Met Office. Yahoo had something like a 30%
accuracy,
UK Weather.com were about 40% I think and the Met Office came out tops
with
60 something %. I wish I'd kept the figures now. My conclusion was,
it's
best to look out of the window :-) And... my goodness are they getting
paid
for this??!

That means that Yahoo was far and away the best forecaster. If you
rigorously stuck to the opposite of what they said, you would have had
70%
accuracy.

Franz

Er... somewhat flawed logic - just because it's right for 30% of the time,
does not mean that the opposite is true for 70% since there are so many
meteorological variables (i.e. the opposite could be just as wrong). What
this thread tells us is that for all the technology and super computer
models being used, the predictions are fundamentally flawed because they
rely on basic physical principles and ignore chaos theory, quantum
mechanics
and the insight of a deity. The suggestion of looking out of one's window
or using the old fashioned techniques (pine cones, 'red sky at night...',
feeling in the bones) may prove just as useful and accurate!

Jason




I wished afterwards that I had predicted the weather myself also, to see how
well I would have done. I was going to do the experiment again and before
starting, I was going to put various "possible weathers" into a hat, then
pull them out and add them in advance to each day before beginning the
experiment with the three websites. I never did it though. I don't think I
would have had much chance of a high score doing it this way, but if I'd
predicted it the day before as the forecasters do *and* listened to my
bones, looked at my pine cone and the sky, then I can't see any reason for
not doing as well as the met office's 60 something %. And then, I could have
set up my own weather website could I not?!

I was quite lenient with my experiment by the way. The met office always
predicted higher temperatures than actually materialised, but I let them off
if the weather was correct. The other two seemed better on temperatures but
would predicted rain when the sun shone and visa versa. I considered this
more seriously wrong than the temperature problems and so this is possibly
why they came out with a lower score.



---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.520 / Virus Database: 318 - Release Date: 18/09/2003