Finite stopping in sequential sampling without recall from a Dirichlet process (Q1077853): Difference between revisions

Let \(X_ 1,X_ 2,...\), be observations from an (unknown) distribution F. The cost of stopping the sequence at n is \(X_ n+(n-1)c\), where c is a constant. The optimal stopping rule is that minimizing the expected cost. The main result states that if F comes from a Dirichlet process than there exists an integer N such that any optimal procedure stops sampling not later than at N with probability 1.