Using Influence Diagrams to Solve the Calibration Problem.
Abstract
The purpose of the calibration experiment is to learn about (alpha beta) so that given a future observation y we can reduce our uncertainty about a future true measurement x. To calibrate our measuring instrument, we record n measurements y = (y sub 1, y sub 2, ..., y sub n) on n units all of whose true measurements, x = (x sub 1, x sub 2, ...,x sub n) are specified beforehand. Based on our prior , p(x), for a future x (call it sub f) and our regression model (1.1), our problem is to determine x = x sub 1, x sub 2,...,x sub n) (subject to feasibility constraints) so as to minimize some overall loss function. x is called the experimental design for the calibration experiment. Keywords: bivariate analysis; normal distribution.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1986
- Accession Number
- ADA176387
Entities
People
- N. G. Smiriga
- R. W. Mensing
- Richard E. Barlow
Organizations
- University of California, Berkeley