Cox, M G; Harris, P M (2015) Polynomial calibration functions revisited: numerical and statistical issues. Advances in Mathematics for Applied Sciences, 86. pp. 9-16.

Full text not available from this repository.

Abstract

The construction of a polynomial calibration function is revisited, paying attention to the representation of polynomials and the selection of an appropriate degree. It is noted that the monomial representation (powers of the `natural' variable) is inferior to the use of monomials in a normalized variable, which in turn is bettered by a Chebyshev representation, use of which also gives stability and insight. Traditional methods of selecting a degree do not take fully into account the mutual dependence of the statistical tests involved. We discuss degree selection principles that are more appropriate.

Item Type:	Article
Keywords:	polynomial calibration, VIM, numerical statistical, uncertainty
Subjects:	Mathematics and Scientific Computing Mathematics and Scientific Computing > Measurement Uncertainties
Last Modified:	02 Feb 2018 13:13
URI:	https://eprintspublications.npl.co.uk/id/eprint/6703