In the case where one is interested in making a probability
statement about the distribution of the random variable of
interest, given that one has observed realisations
one can use the marginal distribution
which is termed the predictive distribution. In practice, this
integral can not generally be calculated, since the analytical
form of 
is not known. However, samples may be
drawn from , in which case the predictive
distribution, together with any other distributions may be
estimated using the kernel density estimate.