Model averaging

A model averaged estimator is composed of estimators, each obtained from a different model, that are weighted and summed. There are different ways of setting the weights. The weights can be deterministic, or random. Frequentist methods for the weights may be based on the values of an information criterion, or may be the result of some optimization procedure. Fully Bayesian methods for model averaging incorporate prior information on the models and the parameters therein. Model averaging is used to avoid choosing a single estimator or to improve individual estimators performances.edition: epubstatus: publishe