Estimating the parameters of a doubly truncated normal distribution
作者:
Mukul M Mittal,
Ram C Dahiya,
期刊:
Communications in Statistics - Simulation and Computation
(Taylor Available online 1987)
卷期:
Volume 16,
issue 1
页码: 141-159
ISSN:0361-0918
年代: 1987
DOI:10.1080/03610918708812582
出版商: Marcel Dekker, Inc.
关键词: maximum likelihood estimation;Bayes' modal estimation nonexistence of the estimator
数据来源: Taylor
摘要:
This paper deals with the maximum likelihood estimation of parameters for a doubly truncated normal distribution when the truncation points are known. We prove, in this case, that the MLEs are nonexistent (become infinite) with positive probability. For estimators that exist with probability one, the class of Bayes modal estimators or modified maximum likelihood estimators is explored. Another useful estimating procedure, called mixed estimation, is proposed. Simulations compare the behavior of the MLEs, the modified MLEs, and the mixed estimators which reveal that the MLE, in addition to being nonexistent with positive probability, behaves poorly near the upper boundary of the interval of its existence. The modified MLEs and the mixed estimators are seen to be remarkably better than the MLE
点击下载:
PDF (487KB)
返 回