TWO-ACTION COMPOUND DECISION PROBLEMS.
Abstract
Asymptotically optimal methods are given for deciding between two alternative actions in situations such as those occurring in routine lot by lot acceptance sampling where essentially the same decision problem arises repeatedly and accumulated past information can be used to advantage. A linear cost structure is used and no assumptions are made concerning the existence of a priori probability distributions of the relevant unknown parameters. Procedures are given which take advantage of empirical dependencies appearing in the sequences of unknown parameter values. Rates of convergence to optimality are obtained. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 11, 1966
- Accession Number
- AD0632535
Entities
People
- M. V. Johns Jr.
Organizations
- Stanford University