The linear calibration problem: a Bayesian analysis

Abstract

The statistical linear calibration problem (or inverse linear regression problem) is the problem of determining a linear relationship between a set of observed data and their corresponding true values and then to be able to predict true values, based on future observed values. The problem is formulated and a variety of solutions are examined from both a frequentist approach and a Bayesian approach with more emphasis on the Bayesian approach. To illustrate the problem, an analysis of a recent archaeological problem, known as the luminescence problem, is examined. The paper concludes with a consideration of design issues and a more general analysis of the estimators.

Publication
MMath Dissertation, University of Durham
Date
Links