Model Variations in Credibility Theory.
Abstract
Classical credibility theory is a linearized Bayesian forecasting method, in which the expected value of the next observation is obtained as a mixture of a prior mean and the sample mean of observed data. In this paper, the author considers a variety of extensions to this model which appear to have practical importance. The main tool is least-squares theory.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1974
- Accession Number
- ADA001338
Entities
People
- William S. Jewell
Organizations
- University of California, Berkeley