Inference for Mean Residual Life and Proportional Mean
Residual Life Model via Empirical LikelihoodIn addition to the distribution function, the mean residual life
Georgia State University
function is the other important function which can be used to
characterize a lifetime.
For inference on the mean residual life function, there is a
standard procedure based on normal approximation. However, the
accuracy of such procedures can be quite low when the censoring
proportion is high. In this talk, we apply an empirical likelihood
(EL) method and derive its limiting distribution. Based on the
result, we obtain a confidence interval for the mean residual life.
Built on mean residual life function, the proportional mean residual
life model with right censoring has been proposed and the semiparametric
inference procedure was investigated. We apply EL to derive its limiting
distribution and obtain a confidence region for the regression parameter.
We compare the proposed methods with that normal approximation based
methods through simulation study. We illustrated the proposed methods
with a real data example.