Yn erthygl , sgrifennodd
Alastair McDonald :
How can you verify a forecast like that? 60% / 40% is very like "Heads
I win. Tails you loose."
As a one-off forecast, yes. But as one of many forecasts, no.
If you have, say, 10 or more such forecasts, and the accuracy is actually
around 20%, then you can say they weren't very good forecasts. Arguably,
you could also say the forecasts weren't good if the accuracy was 90%
(at least, the probability estimating wasn't good!).
Adrian
--
Adrian Shaw ais@
Adran Cyfrifiadureg, Prifysgol Cymru, aber.
Aberystwyth, Ceredigion, Cymru ac.
http://users.aber.ac.uk/ais uk