CHAPTER 9
Advanced Topics in Estimation Theory
PART I: THEORY
In the previous chapters, we discussed various classes of estimators, which attain certain optimality criteria, like minimum variance unbiased estimators (MVUE), asymptotic optimality of maximum likelihood estimators (MLEs), minimum mean–squared–error (MSE) equivariant estimators, Bayesian estimators, etc. In this chapter, we present additional criteria of optimality derived from the general statistical decision theory. We start with the game theoretic criterion of minimaxity and present some results on minimax estimators. We then proceed to discuss minimum risk equivariant and standard estimators. We discuss the notion of admissibility and present some results of Stein on the inadmissibility of some classical estimators. These examples lead to the so–called Stein–type and Shrinkage estimators.
9.1 MINIMAX ESTIMATORS
Given a class of estimators, the risk function associated with each d is R(d, θ), θ Θ. The maximal risk associated with d is R*(d) = R(d, θ). If in there is an estimator d* that minimizes R*(d) then d* is ...
Get Examples and Problems in Mathematical Statistics now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.