scientific article; zbMATH DE number 23007
Let \(X_ 1,X_ 2,\dots\) be i.i.d. r.v.'s following the density \(f(x;\theta)\), where \(\theta\) is an unknown parameter. Suppose that for the purpose of estimation of \(\theta\) we have at our disposal only a sample of random size \(X_ 1,\dots,X_{\nu_ p}\), where \(\nu_ p\) is distributed according to the geometric distribution with the parameter \(p\), i.e., \[ P(\nu_ p=k)=p(1-p)^{k-1}, \quad k=1,2,\dots, \quad 0<p<1. \] The main aim is to estimate \(\theta\) by means of the maximum likelihood method and to study the properties of such an estimator. It is shown that for the corresponding likelihood equation \[ \nu_ p^{- 1}\sum_{i=1}^{\nu_ p}(\partial/\partial\theta)\log f(X_ j,\theta)=0, \] there exists a solution \(\hat\theta_ p\) and \(\hat\theta_ p\to\theta\) as \(p\to 0\) in probability. Moreover, the asymptotic distribution of \(p^{-1/2}(\hat\theta_ p-\theta)\) is derived.
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3980804)
