using R to estimate finite mixture model with underlying Markov process

Question

My apologies if this is more of a statistics question than an R question. I am trying to estimate the following model in R.

y_t = mu0 (1 - S_t) + mu1 S_t + e_t e_t ~ N(0, sigma_t^2) sigma_t^2 = sigma_0^2 (1 - S_t) + sigma_1^2 S_t

where mu_t = mu0 if S_t = 0, mu_t = mu1 if S_t = 1, and S_t is a Markov process, either 0 or 1, with transition probabilities P(S_t = 1 | S_t-1 = 1 ) = p and P(S_t = 0 | S_t-1 = 0 ) = q.

Would 'flexmix' be a good library to use for this? I am new to this kind of statistics so any pointer to the right library would be appreciated.

Thanks,

Answer 1

This looks like the exactly type of model you could easily code up in Bugs or Jags. Bugs/Jags is probably the most flexible approach to estimating custom models in R. You can easily move between R and Jags using R2Jags.

If you are new to Bayesian models, it may take a bit to get up to speed.