Overview
Comparison of measurement results, reliable decision-making and conformity assessment require the evaluation of uncertainties associated with measurement results. The ability to compare measurements made in different places and at different times underpins international metrology. The Guide to the Expression of Uncertainty in Measurement (GUM) provides guidance for the evaluation of uncertainties, and it has been applied successfully in many applications throughout metrology.
Illustration of Monte Carlo method according to Supplement 1 to the GUM.
In recent years metrology has expanded to support new fields to address societal challenges relating to energy and sustainability, climate and environmental monitoring, life sciences and health, using measurement modalities such as imaging, spectroscopy, earth observation and sensor networks. Reliable uncertainty evaluation is particularly important in these applications, e.g. to safeguard the diagnosis of a tumor in quantitative imaging or to reliably monitor air pollution. The GUM does not adequately address the challenges arising in these applications, and the development of statistical procedures for improved uncertainty evaluation is an urgent need.
Research
The focus of PTB’s Working Group 8.42 is on the development of Bayesian methods for the evaluation of uncertainties. The development is carried out within the context of different research areas of data analysis such as large-scale data analysis or deep learning. Bayesian inference procedures suitable for the extension of the current GUM methodology are also part of the current research in PTB’s Working Group 8.42. Examples include simple means to assign distributions representing the available prior knowledge, or procedures for the numerical calculation of results. Open source software support is provided to ease the application of the research results.
Software
Publications
Publication single view
Article
Title: | A tutorial on Bayesian Normal linear regression |
---|---|
Author(s): | K. Klauenberg, G. Wübbeler, B. Mickan, P. Harris and C. Elster |
Journal: | Metrologia |
Year: | 2015 |
Volume: | 52 |
Issue: | 6 |
Pages: | 878--892 |
DOI: | 10.1088/0026-1394/52/6/878 |
Tags: | 8.42, Regression, Unsicherheit |
Abstract: | Regression is a common task in metrology and often applied to calibrate instruments, evaluate inter-laboratory comparisons or determine fundamental constants, for example. Yet, a regression model cannot be uniquely formulated as a measurement function, and consequently the Guide to the Expression of Uncertainty in Measurement (GUM) and its supplements are not applicable directly. Bayesian inference, however, is well suited to regression tasks, and has the advantage of accounting for additional a priori information, which typically robustifies analyses. Furthermore, it is anticipated that future revisions of the GUM shall also embrace the Bayesian view.Guidance on Bayesian inference for regression tasks is largely lacking in metrology. For linear regression models with Gaussian measurement errors this tutorial gives explicit guidance. Divided into three steps, the tutorial first illustrates how a priori knowledge, which is available from previous experiments, can be translated into prior distributions from a specific class. These prior distributions have the advantage of yielding analytical, closed form results, thus avoiding the need to apply numerical methods such as Markov Chain Monte Carlo. Secondly, formulas for the posterior results are given, explained and illustrated, and software implementations are provided. In the third step, Bayesian tools are used to assess the assumptions behind the suggested approach.These three steps (prior elicitation, posterior calculation, and robustness to prior uncertainty and model adequacy) are critical to Bayesian inference. The general guidance given here for Normal linear regression tasks is accompanied by a simple, but real-world, metrological example. The calibration of a flow device serves as a running example and illustrates the three steps. It is shown that prior knowledge from previous calibrations of the same sonic nozzle enables robust predictions even for extrapolations. |