Sequential selection procedures. A decision theoretic approach (Q794084)

Let \(\pi_ 1,\pi_ 2,...,\pi_ k\) be given populations which are associated with unknown real parameters \(\theta_ 1,\theta_ 2,...,\theta_ k\) from a common underlying exponential family \({\mathcal F}\). Let the goal be to select a subset (of random or fixed size) of good populations (i.e. those which have large parameters). In this paper, the authors have studied sequential procedures for this problem in a general framework covering the control and noncontrol, elimination and nonelimination, truncated and open sequential settings. They have derived their results under a Bayesian approach with symmetric priors and have used the concepts of backward induction and decrease in transposition.