Information matrix for a mixture of two inverse gaussian distributions
作者:
E. K. AL-Hussaini,
K. E. Ahmad,
期刊:
Communications in Statistics - Simulation and Computation
(Taylor Available online 1984)
卷期:
Volume 13,
issue 6
页码: 785-800
ISSN:0361-0918
年代: 1984
DOI:10.1080/03610918408812415
出版商: Marcel Dekker, Inc.
关键词: identifiability of finite mixtures;power series expansion
数据来源: Taylor
摘要:
In this paper, Fisher information matrix about the five parameters ρ, μ:1, μ2, λ1and λ2of a mixture of two Inverse Gaussian density functions is obtained. The Leguerre-Gauss quadrature formula is used to evaluate the essential integral on which the twenty five elements of the information matrix are based. Results involving the computation of the information about p are compared with those involving both the power series expansion and Simpson's method of integration. Laguerre-Gauss quadra-ture was found to lead to good approximations as compared with other methods. It was therefore chosen for the computations of the elements of the information matrix.
点击下载:
PDF (452KB)
返 回