Bayesian updating of prior distributions

All prior distributions described in Prior can be updated with binomial observations (x out of n responses).

# S4 method for PointMass
update(object, x, n)

# S4 method for Beta
update(object, x, n)

# S4 method for BetaMixture
update(object, x, n)

# S4 method for GenericDistribution
update(object, x, n)

# S4 method for JeffreysPrior
update(object, x, n)

Arguments

object	Prior distribution object to update
x	number of responses out of n individuals
n	number of individuals out of which x had a response

See also

Prior, condition

Examples

if (FALSE) {
  # point mass distributions are invariant under updating
  update(PointMass(.4), 3, 10)
}

if (FALSE) {
  update(Beta(5, 7), 3, 10) # same as Beta(8, 14)
}

if (FALSE) {
  update(1/3*Beta(5, 7) + 2/3*Beta(1,1), 3, 10) # update mixtures
}
if (FALSE) {
  design    <- Design(c(0, 30, 25, 0), c(Inf, 10, 7, -Inf))
  prior     <- JeffreysPrior(design)
  posterior <- update(prior, 3, 10) # results in a GenericDistribution object (no analytical update)
  update(posterior, 2, 5) # the generic posterior of a Jeffreys prior can also be updated again
}