posterior: Tools for Working with Posterior Distributions in R

Le résumé fourni par la source

Modern Bayesian inference is often performed via sampling algorithms that produce draws (samples) from the model's posterior distribution.The most important class of such algorithms is Markov chain Monte Carlo (MCMC), but also other algorithm classes such as variational inference and neural posterior estimation rely on posterior draws as their primary output representation.Regardless of their specific origin, these draws have to be stored and postprocessed to obtain insights into the Bayesian inference results.In this context, relevant questions for the Bayesian modeler include in which format to store the posterior draws, which diagnostics to run to assess the trustworthiness of the obtained draws, and how to best summarize the draws for inference and decision-making.Due to the widespread use of sampling algorithms in Bayesian inference, essentially all Bayesian modelers face these questions during their data analyses, thus strongly benefiting from modern, efficient, and easy-to-use solutions.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.