Quality of p(
q
)
n
Use the expected loss of the Bayesian point
estimate as a
measure of quality
of p(
q
):
n
n
Risk(p) =
n
p(
q
)
q
~
p(
q
)
q
~