ABSTRACT
Markov Chain Monte Carlo (MCMC) methods encompass a general framework of methods introduced by Metropolis et al. and Hastings for Monte Carlo integration. The MCMC estimate of the posterior mode, percentile interval, and highest posterior density interval for the derived parameter is computed. The estimates of the variance of the statistic are analogous to estimates based on between-sample and within-sample mean squared errors in a one-way analysis of variance. The coda package provides utilities that summarize, plot, and diagnose convergence of mcmc objects created by functions in MCMC pack. Convergence of the random walk Metropolis is often sensitive to the choice of scale parameter. The random walk Metropolis sampler is an example of a Metropolis sampler. If the variance of the increment is too small, the candidate points are almost all accepted, so the random walk Metropolis generates a chain that is almost like a true random walk, which is also inefficient.
