Correction re. Jeffreys' pseudocounts

This commit is contained in:
hyginn 2017-11-20 19:00:23 -05:00
parent 9a987c22be
commit a1b5eb7b90

View File

@ -482,7 +482,7 @@ pmf
# for ordered data one could substitute the average values of the two bracketing
# outcomes. But a simple and quite robust solution is to add "pseudocounts".
# This is called adding a Laplace prior, or a Jeffreys prior: in our case,
# simply add 0.5 to every value that the two functions don't share.
# simply add 0.5 to every category.
# pmf of an honest die
pmfHD <- rep(1/6, 6)