A package eglhmm ("extended generalised linear hidden Markov models")
has just become available on CRAN. Mac OSX binaries are available.
Windoze binaries will presumably be available shortly. The new package
largely supersedes the package hmm.discnp (which will eventually be
archived). The hmm.discnp package deals only with ad hoc discrete
distributions, which are handled in eglhmm by specifying the
distribution to be "Multinom". The eglhmm package is capable of
handling Gaussian, Poisson, Binomial, and discretised beta (Dbd)
distributions as well as the Multinomial distribution.
I would appreciate learning about any problems or bugs that users
encounter in using eglhmm. General comments and suggestions are also
welcome.
Note that eglhmm has considerable overlap with the depmixS4 package
(written by Ingmar Visser and Maarten Speekenbrink), but a considerable
symmetric difference as well. The syntax is very different.
cheers,
Rolf Turner
--
Honorary Research Fellow
Department of Statistics
University of Auckland
Stats. Dep't. (secretaries) phone:
+64-9-373-7599 ext. 89622
Home phone: +64-9-480-4619