Affordable Access

-deficiency of the Kaplan-Meier estimator



Let X1,...,Xn,... be a sequence of independent and identically distributed random variables with distribution function F subject to random right censoring. Considering the classical Kaplan-Meier estimator and a smoothed kernel-type estimate , we prove that and (mean integrated absolute error) tend to the same constant as n goes to infinity. However, we establish that the smoothed estimator has a performance better than (for some bandwidths) what relative -deficiency is of interest. The optimal choice of the bandwidth hn, with respect to MIAE sense, is also obtained.

There are no comments yet on this publication. Be the first to share your thoughts.