Out of a sample of 1000 people, 2 have cancer. If for each person you guess randomly (half the time you guess cancer, half the time no cancer), how accurate will you be on average. Aside from just guessing that everyone doesn't have cancer, is there a better methodology in guessing that would yield a higher accuracy?

Question

Out of a sample of 1000 people, 2 have cancer.  If for each person you guess randomly (half the time you guess cancer, half the time no cancer), how accurate will you be on average.  Aside from just guessing that everyone doesn't have cancer, is there a better methodology in guessing that would yield a higher accuracy?

phi · Answer

you could draw a tree diagram
you pick a person, and they have C with Pr(C)
and (not C) with Pr( not C)
then make a guess 1/2 C and 1/2 not C
|dw:1475339386539:dw|