Evaluation of cycles of comparison measurements by a least-squares method
Ignacio H Lira · Measurement Science and Technology · 2001
This paper discusses the least-squares procedure for evaluating comparison measurements. The input information consists of a series of indicated values of at least two quantities. The measurands are the differences between the quantities involved. Two algorithms are presented in matrix notation. They can handle the general case in which measurements are correlated. Attention is also paid to independent equal-variance measurements, to the case in which temporal drift occurs and to the consistency of the least-squares adjustment. An example illustrates the application of both algorithms.