Bayes on holiday
July 2009 (index)
tourists?
My first reaction on seeing this gentleman today is that he must be on holiday.
At a rough guess something like 10% of the people in that area were on holiday. Maybe 1% of men on holiday would would dress like that; but only 1 in 10,000 locals. We'll write:
H for being on holiday
~H for being not on holiday
C for wearing those clothes
Then:
P(H|C) = P(C|H)P(H) ÷ P(C) (Bayes)
P(C) = P(C|H)P(H) + P(C|~H)P(~H)
= 0.01*0.1 + 0.0001*0.9
= 0.001 + 0.00009
= 0.00109
P(H|C) = P(C|H)P(H) ÷ P(C)
= 0.001 ÷ 0.00109
= 92%
If anything 92% feels a bit low. If that's the case then one of our original estimates must be wrong. I suspect the proportion of locals who dress like that is even lower than 1 in 10,000. Perhaps nil.
ibbly.com contact