A Method for Constructing a Unimodal Inferential or Prior Distribution.
TEXAS UNIV AT AUSTIN CENTER FOR CYBERNETIC STUDIES
Pagination or Media Count:
This paper shows how to take personally assessed information and use it to develop a continuous unimodal prior density function, perhaps for subsequent Bayesian analysis. The method is completely nonparametric and uses only the furnished information and no other. The technique is easily computerized, and yields a closed analytical formula for the prior. The resulting distribution may be considered to be an inferential distribution. Author
- Statistics and Probability