Nonparametric hazard rate estimation with associated kernels and minimax bandwidth choice

In this paper, we consider the general theory of nonparametric hazard rate estimation with associated kernels, for which the shape of the kernel depends on the point of estimation. We prove MISE convergence results and a central limit theorem for such estimators. We then prove an oracle type inequality for both a local and global minimax bandwidth choice. The results are then illustrated by showing that they apply to the Gamma kernel and providing numerical simulations and an application to biological data.