Hazard rate function estimation using inverse gaussian kernel

Abstract EN

In this paper, we consider the nonparametric estimation of the hazard rate function for independent identically distributed (iid) data using kernel estimation techniques.

Since survival times are positive with potentially a high concentration at zero, one has to take into account the bias problem when the hazard rate function is estimated in the boundary region.

To overcome the boundary bias problem, we use the Inverse Gaussian (IG) kernel, since it has a positive support.

The asymptotic mean squared error (AMSE) and the asymptotic normality of the proposed estimator are investigated.

Also, the selection of an optimal bandwidth is discussed since it plays an important role in the kernel estimation.