两参数广义指数分布的参数估计与数值模拟-应用统计专业论文.docxVIP

华中科技大学硕士学位论文 华 中 科 技 大 学 硕 士 学 位 论 文 II II Abstract Since Gupta and Kundu argued two-parameter generalized exponential (GE) distribution, it has caused the attention of many scholars and they research it widely. This distribution can better fit some life data, and is widely used in many fields .Many scholars study on the parameter estimation and its property of GE distribution , but they mainly study in a single case under the same loss function .Under different loss functions ,the parameter estimation of GE distribution is less discussed ,especially the study on the E-Bayes estimation and hierarchical Bayes estimation is very few. On the basis of predecessors research, this paper introduce GE distribution in the whole sample type censored-I data model , we consider different situations, with different loss functions, discuss E-Bayes estimation and hierarchical Bayes estimation, and with the Monte Carlo simulation. In the first chapter,we introduce the present situation, the model and the main work. In the second chapter, we introduce the theory of Bayes statistics , the related theorem and definitions. In the third chapter, when λ is known, we give the MLE , the Bayes estimation、 E-Bayes estimation and hierarchical Bayes estimation of α and R under two different loss functions, with MC simulation we find that the error of the parameter estimation is smaller under LINEX loss function, and E-Bayes estimation and hierarchical Bayes estimation effect well. In the fourth chapter, when α and λ are unknown , the MLE of α and λ are given, with Lindley approximation theorem we get the Bayes estimation of α and λ , in the last section , through the MC simulation, we find the effect is better under the square loss function. In the end ,the full text is summarized, we point out the merit and shortcoming and we further prospect the paper. Key words: Generalized exponential distribution; Type censored-I data; Parameter estimation; Square loss function; LINEX loss function