equivalent identity for posterior predictive distribution

16 Views Asked by At

Let $x ~ p(x)$ be I.I.D.

The posterior predictive distribution for the predicted observations based on the realised observations is

$p(x' | x) = \int d\theta p(x' | x, \theta) p(\theta | x) = \int d\theta p(x' | \theta) p(\theta | x)$

How does the second equality relies on the IID assumption?