Premium
Estimating the bivariate mean vector of censored environmental data with Box–Cox transformations and E‐M algorithm
Author(s) -
Freeman Jade,
Modarres Reza
Publication year - 2006
Publication title -
environmetrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.68
H-Index - 58
eISSN - 1099-095X
pISSN - 1180-4009
DOI - 10.1002/env.778
Subject(s) - mathematics , covariance matrix , statistics , power transform , multivariate normal distribution , bivariate analysis , transformation (genetics) , skew , covariance , data set , estimation of covariance matrices , confidence interval , algorithm , multivariate statistics , computer science , telecommunications , biochemistry , gene , chemistry , geometry , consistency (knowledge bases)
We present a method for estimating the mean vector from a bivariate skew distribution that includes some unobserved data below the detection limits. The method uses a Box‐Cox transformation, of which the parameters are found by maximizing the likelihood function over a fixed power transformation set. To estimate the mean vector and the covariance matrix, we develop an E‐M algorithm solution. Given a transformation, we obtain expressions for the mean vector, covariance matrix, and the asymptotic covariance of the vector of means in the original scale. Expressions are obtained for a confidence region for the vector of means. The performance of the maximum likelihood estimation (MLE) method in selecting the correct power transformation and the coverage rate of the confidence region under several conditions are investigated in a simulation study. This method gives reliable results for finding effective transformations and the coverage rate for highly skew data sets. The method is applied to water quality monitoring data. Copyright © 2005 John Wiley & Sons, Ltd.