Kobkun Raweesawat. Odds ratio estimation in rare event. Doctoral Degree(Applied Statistics). King Mongkut's University of Technology North Bangkok. Central Library. : King Mongkut's University of Technology North Bangkok, 2016.
Abstract:
This thesis is aimed to propose a new estimator of odds ratio in rare events using Empirical Bayes. The study is performed on two distributions ; binomial distribution with informative priors as Beta distribution and Poisson distribution with informative priors as Gamma distribution when data are constructed in two groups with equal and unequal sample sizes : (10,10), (10,30), (30,10) and (10,50). In each group, the probabilities of success are varied to 0.008, 0.01, 0.03, 0.05, 0.10 and 0.15. In each situation, data are simulated with Markov chain simulation algorithm and repeated 5,000 times after 1,000 burn-ins, using R program. The average of percentage of estimated relative error is used as a criterion for comparison. The results indicated that Empirical Bayes (EB), both in binomial and Poisson distributions is more efficient when compared to the modified maximum likelihood estimator (MMLE) and the median unbiased estimator (MMUE).