Smooth Function Estimation

Abstract

Nonparametric estimation for lifetime or continuous data has its roots in the fifteenth century. However, until the 1950s, all nonparametric estimators obtained for the probability density function and the cumulative distribution function were discrete. Typically, these estimates were hard to update and required more computation than necessary. This was not a serious problem if the functions being estimated were themselves discrete, but most lifetime or measurement data are continuous and so “smoother” estimators were needed.