Bayesian model averaging
(1.2 hours to learn)
Summary
In model selection, we typically select a single "best" model from a set of candidate models (based upon some selection criteria, such as an AIC score) and then use this model for prediction. Instead of selecting a single "best" model and using it for prediction, Bayesian Model Averaging BMA uses a weighted average of each model's individual prediction for the final predicted value, where the weight is the posterior probability of the model given the data.
Context
This concept has the prerequisites:
- Bayesian model comparison (Bayesian model averaging is based on the same ideas as Bayesian model comparison.)
Core resources (read/watch one of the following)
-Free-
→ Bayesian Model Selection and Model Averaging
Supplemental resources (the following are optional, but you may find them useful)
-Free-
→ Bayesian Model Averaging: A Tutorial
Location:
Sections 1-3 provide core material and sections 4-7 provide examples and further technical details
See also
- Reversible jump MCMC is a class of sampling algorithms often used for Bayesian model averaging.