Boom: suggest.burn.log.likelihood – R documentation

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

suggest.burn.log.likelihood

Suggest MCMC Burn-in from Log Likelihood

Description

Suggests a burn-in period for an MCMC chain based on the log likelihood values simulated in the final few iterations of the chain.

Usage

SuggestBurnLogLikelihood(log.likelihood, fraction = .10, quantile = .9)

Arguments

`log.likelihood`	A numeric vector giving the log likelihood values for each MCMC iteration.
`fraction`	The fraction of the chain that should be used to determine the log likelihood lower bound. The default setting looks in the final 25% of the MCMC run. Must be an number less than 1. If `fraction <= 0` then a 0 burn-in is returned.
`quantile`	The quantile of the values in the final fraction that must be exceeded before the burn-in period is declared over.

Details

Looks at the last fraction of the log.likelihood sequence and finds a specified quantile to use as a threshold. Returns the first iteration where log.likelihood exceeds this threshold.

Value

Returns a suggested number of iterations to discard. This can be 0 if fraction == 0, which is viewed as a signal that no burn-in is desired.

Author(s)

Steven L. Scott steve.the.bayesian@gmail.com

Boom

Bayesian Object Oriented Modeling

v0.9.7

LGPL-2.1 | file LICENSE

Authors

Steven L. Scott is the sole author and creator of the BOOM project. Some code in the BOOM libraries has been modified from other open source projects. These include Cephes (obtained from Netlib, written by Stephen L. Moshier), NEWUOA (M.J.D Powell, obtained from Powell's web site), and a modified version of the R math libraries (R core development team). Original copyright notices have been maintained in all source files. In these cases, copyright claimed by Steven L. Scott is limited to modifications made to the original code. Google claims copyright for code written while Steven L. Scott was employed at Google from 2008 - 2018, but BOOM is not an officially supported Google project.

Initial release

2021-02-15