MINIMUM MSE ESTIMATION BY LINEAR COMBINATIONS OF ORDER STATISTICS1

From MaRDI portal





In this paper we assume that in a random sample of size n drawn from a population having the pdf f(x;\(\theta)\) the smallest \(r_ 1\) observations and the largest \(r_ 2\) observations are censored \((r_ 1\geq 0\), \(r_ 2\geq 0)\). We consider the problem of estimating \(\theta\) on the basis of the middle \(n-r_ 1-r_ 2\) observations when either \(f(x;\theta)=\theta^{-1}f(x/\theta)\) or \(f(x;\theta)=(a\theta)^{- 1}f((x-\theta)/a\theta)\) where f(\(\cdot)\) is a known pdf, a \((>0)\) is known and \(\theta\) \((>0)\) is unknown. The minimum mean square error linear estimator of \(\theta\) proposed in this paper is a shrinkage of the minimum variance linear unbiased estimator of \(\theta\). We obtain explicit expressions of these estimators and their mean square errors when (i) f(\(\cdot)\) is the uniform pdf defined on an interval of length one and (ii) f(\(\cdot)\) is the standard exponential pdf, i.e., \(f(x)=\exp (-x)\), \(x\geq 0\). Various special cases of censoring from the left (right) and no censoring are considered.











This page was built for publication: MINIMUM MSE ESTIMATION BY LINEAR COMBINATIONS OF ORDER STATISTICS1

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3684992)