Minimum Hellinger distance estimation of parameter in the random censorship model (Q1175378)

Let \(X_ 1,\dots,X_ n\) be i.i.d. random variables with lifetime c.d.f. \(F\) on \([0,\infty)\) and \(Y_ 1,\dots,Y_ n\) be independent of \(X_ i\)'s and i.i.d. with censoring c.d.f. \(G\) on \((0,\infty)\). The author considers the random censorship model, when the pairs \(\{\min(X_ i,Y_ i)\mid[X_ i\leq Y_ i]\}\), \(1\leq i\leq n\), are observed. Suppose that \(F\) has a density \(f\) with respect to the Lebesgue measure, and some physical theory suggests that \(f\) belongs to a parametric family \(\{f_ \theta\mid\theta\in\Theta\}\) where \(\Theta\subseteq R_ p\). At the same time the author assumes that due to a variety of data contamination, f may possibly differ from any of the \(f_ \theta\)'s. The problem is to estimate the parameter that gives the ''best fit'' of the parametric model to the data. Minimum Hellinger distance estimation is considered. At first some preliminary results are introduced. The tail behavior of the product limit process is investigated and the weak convergence of the process is established on the entire support set. The convergence of the kernel density estimator in the Hellinger metric is obtained. In addition, an upper bound on the mean square increment of the normalized \(P-L\) process is developed. Also, asymptotic behavior of MHDE is investigated and its robustness properties are briefly discussed and results of simple examples are presented.