Adaptive Robust Estimation Of Location And Scale Parameters Using Selected Discriminants

Author : Mercidita Tulay Navarro
Publisher :
ISBN 13 :
Total Pages : 212 pages
Book Rating : 4.:/5 (244 download)

DOWNLOAD NOW!

Download or read book Sequential Estimation of Quantiles and Adaptive Sequential Estimation of Location and Scale Parameters written by Mercidita Tulay Navarro and published by . This book was released on 1990 with total page 212 pages. Available in PDF, EPUB and Kindle. Book excerpt: Suppose that $Xsb1,Xsb2,cdots$ are independent observations from a distribution F, and that one wishes to estimate the $psp{rm th}$ quantile $xisb{p}(0 0).$ If $f(xisb{p})$ is known, one may use the best fixed sample size (i.e., in the sense of minimum risk). If $f(xisb{p})$ is unknown, then the best fixed sample size is also unknown. For this case, a stopping rule T = $Tsb{A}$ is proposed. It is shown that, under certain smoothness conditions on F and a growth condition on the delay, the sequential procedure derived is asymptotically risk efficient, i.e., it performs asymptotically as well as the best fixed-sample-size procedure. In the proof, asymptotic moment bounds for the remainder terms in Bahadur's representations of sample quantiles, and certain central order statistics, are derived and used to verify uniform integrability of $$left{left(Asp{1/4}vert xisb{pT} - xisb{p}vertright)sp2, A geq 1right}.$$Results are extended to a problem that utilizes a more general loss function, $Lsb{n}prime = Avert xisb{pn} - xisb{p}vertsp{r} + n (A > 0, r > 0),$ and to estimation of a linear combination of two quantiles. The problem of estimating the center of a symmetric distribution using the better of the sample mean and the sample median, and that of estimating scale using the better of the sample standard deviation and the sample interquartile range, are studied. Under certain conditions, the risk is minimized (asymptotically) by using the estimator that possesses the smaller asymptotic variance (in scale estimation, the smaller standardized asymptotic variance) and the sample size that minimizes the asymptotic risk of that estimator. Oftentimes, though, the asymptotic variances of the estimators are unknown. In this case, no single estimator and no fixed sample size would minimize the risk. Adaptive sequential procedures that allow one to choose not only the estimators but also the sample sizes in the two problems are investigated. Both interval and point estimation are considered.