ESTIMASI PARAMETER REGRESI POISSON MENGUNAKAN ALGORITMA METROPOLIS HASTING; ESTIMATING THE PARAMETER OF POISSON REGRESSION USING METROPOLIS HASTING ALGORITHM

DIYANI ARIF SETYORINI, Subanar

2014 | Disertasi | PROGRAM STUDI S2 MATEMATIKA

Regression analysis is a statistical method used to analyze the relationship between a dependent variable and one or more independent variables. Regression analysis assumed that the dependent variable the distribution Poisson is called Poisson regression analysis. Poisson regression analysis was used to analyze the relationship between a dependent variable stating Poisson distributed discrete data with one or more independent variables. In general, a Poisson regression parameter estimation using classical methods based only on current information obtained from the sample without taking into account the initial information from the Poisson regression parameters. If owned preliminary information about the parameters of the prior distribution, the estimated parameters can use Bayes methods. Bayesian methods combine information on the sample prior to the distribution of information, and the results are expressed in the posterior distribution. If the posterior distribution can not be derived analytically then approximated using Markov Chain Monte Carlo algorithm (MCMC) especially Metropolis Hastings algorithm. This algorithm uses acceptance and rejection mechanism for generating a random sample sequence.

Kata Kunci : Poisson regression, Bayesian methods, Markov Chain Monte Carlo (MCMC), the Metropolis-Hastings algorithm.