Does anyone have any examples of producing the expectation of the posterior predictive (i.e. what brms does with epred functions) in Stan (presumably in the generated quantities block)? The User’s Guide has good examples of generating the posterior predictive distribution of \tilde{y} from new predictor data \tilde{x}. However, this is the posterior of \tilde{y} and not the posterior of E[\tilde{y}].
I imagine an ideal example would show how to predict that quantity for a single data new data point from a relatively simple hierarchical model.