Lesson 18: Bayesian Model Selection скачать в хорошем качестве

Lesson 18: Bayesian Model Selection

Boston University EE509 "Applied Environmental Statistics" Course: In this lecture we review frequentist model selection metrics and discuss the idea of statistical Power, including Monte Carlo simulation methods for estimating power. We then dive into Bayesian model selection metrics, including Deviance Information Criteria (DIC), Watanabe-Akaike (WAIC), and Predictive Loss, including demonstrating how to calculate these in JAGS. Finally, we briefly touch on Bayes Factors, Reversible-jump MCMC, Bayesian Model averaging, and the broader framework introduced by Hooten & Hobbs 2015 "A guide to Bayesian model selection for ecologists" Ecological Monographs