multivariate poisson hidden markov models for analysis of spatial ...

Step 2: Then take a bootstrap sample (with replacement) of size equal to the original
sample size and estimate the parameters using the EM algorithm.
Step 3: Take at least 100 bootstrap samples and estimate the parameters.
Step 4: Finally using these 100 bootstrap parameters calculates the standard errors of the
estimates.
5.6 Splus/R codes for the multivariate Poisson hidden Markov model
We contributed to the hidden Markov model research area by developing Splus/R codes
for the analysis of the multivariate Poisson hidden Markov model. Splus/R codes are
written to estimate the multivariate Poisson hidden Markov model using the EM
algorithm and the forward-backward procedure and the estimation of bootstrapped
standard errors. The estimated parameters were used to calculate the goodness of fit
measures mention in this thesis: the entropy criterion (section 6.5) and the estimated
unconditional variance-covariance matrix (section 7.3). Splus/R programs (see
Appendix) of this thesis are available on request from the author.
5.7 Loglinear analysis
Loglinear models were used to identify the covariance structure in this thesis. The
loglinear model is a special case of generalized linear model (GLM) for count-type
response variables modelled as Poisson data (Agresti, 2002). All generalized linear
models have three components. The random component identifies the response variable
102