Comment on article by Rydén
Sylvia Frühwirth‐Schnatter · Bayesian Analysis · 2008
First of all, I would like to congratulate the author on an excellent paper providing a very fair comparison of EM versus MCMC estimation for hidden Markov models. I agree with most of what has been said in the paper, nevertheless I would like to comment on several issues, including estimating the unknown number of states using marginal likelihoods, choosing the prior and post-processing the MCMC draws to deal with label switching. Before doing so, I would like to introduce bayesf, a MATLAB software package for estimating hidden Markov models.