Abstarct: Markov chain Monte Carlo (MCMC) algorithms, such as Metropolis-Hastings algorithm and Gibbs sampling, are methods for producing samples from complex probability distributions. If the speed of convergence is slowed down in the MCMC methods, the algorithm runs for a long time. One of the reasons that make the convergence of the MCMC algorithms to be slow is separately updating the components of the distribution that are dependent. Group blocking of elements that are dependent is one of the conventional approaches to avoid such a problem. In this thesis, we represent an automated blocking method with a good performance to prevent subjective blocking. Finally, we evaluate the automated blocking process by using both real data and simulated examples and compare the results with those from standard MCMC algorithms.