Bayesian parameter estimation in exponential families

(30 minutes to learn)


Exponential families are convenient for Bayesian parameter estimation because the conjugate priors often have a convenient form, and there is a simple form for the posterior.


This concept has the prerequisites:


  • How do you derive the conjugate prior for an exponential family distribution?
  • Show that the posterior can be computed in terms of the sufficient statistics.
  • Work through a simple example, such as the beta-Bernoulli model.

Core resources (read/watch one of the following)


See also

-No Additional Notes-