From the species problem to a general coverage problem via a new interpretation (Q1193368)

Three problems, apparently different, are studied under the same theoretical framework. They are: i) the structure of the species problem, ii) the problem of estimating the volume of a convex set and iii) the missile-coverage problem. The central idea consists of creating information about the unknown random quantity to be estimated, which depends on the sample \((X_ 1,\dots,X_ n)\), deleting one observation, then calculating the estimators with the \(n-1\) remaining observations and finally averaging those estimators. This methodology provides the explanation for the Turing-Good-Robbins estimator, in the species problem, and it also proves why this estimator is always biased. Moreover, this approach, mixed with an iterative procedure, yields an important bias reduction for the building estimators.