Index: The Book of Statistical ProofsGeneral TheoremsBayesian statisticsProbabilistic modeling ▷ Posterior predictive distribution

Definition: Consider a full probability model with likelihood function $p(y \vert \theta)$ and posterior distribution $p(\theta \vert y)$ based on measured data $y$. Then, the marginal distribution of new data $y_{\mathrm{new}}$, predicted based on the posterior distribution, is called the posterior predictive distribution:

\[\label{eq:post-pred} p(y_{\mathrm{new}} \vert y) = \int p(y_{\mathrm{new}} \vert \theta) \, p(\theta \vert y) \, \mathrm{d}\theta \; .\]
 
Sources:

Metadata: ID: D201 | shortcut: post-pred | author: aloctavodia | date: 2024-08-18, 07:50.