Markov Chain Monte Carlo methods in statistical inference.
Venkatesan, P.; Kannan, K.S.; Vallinayagam, V. Contributions to Statistical Research
Manonmaniam Sundaranar University; 2005; 36-54.
Summary: Markov Chain Monte Carlo (MCMC) methods make possible the use of flexible models that would otherwise be computationally infeasible. In recent years, a great variety of such applications with special reference to Bayesian inference have been described in the literature. Many authors have discussed the effort and expertise needed to design and use of Markov Chain samplers and the confidence one can have on MCMC estimates and its effect on the model building process. This paper reviews some of the recent advances in MCMC that allows exact statistical inference using simulations and identification of models for which good MCMC algorithms exist. The applications of MCMC to both Bayesian and frequentists are discussed with applications to biomedical data.
Keywords: Mixed Model, Gibbs Sampler, Metropolis Algorithm, ECM Algorithm
Back to List of publications / Home