| |
:: Course SummaryThis training session discusses how to measure uncertainty in a measurement process: statistical approach, characterization of a measurement system, types of errors, impact of errors on data quality, accuracy and reliability, analytical approach to measuring uncertainty, propagation of errors on the result of a measurement and how to report results.:: Learning ObjectivesUpon completion of this course, participants will be able to:
Understand the importance of uncertainty analysis
Identify sources of uncertainty and quantify their impact
Master statistical and mathematical tools to measure uncertainty
Learn how to calculate, report and interpret uncertainty measures:: Target AudienceThis applied training session in statistics is aimed at:
Laboratory and quality control staff
Anyone who takes measurements as part of their daily activities:: PrerequisiteThis workshop introduces the important concepts in statistics and data analysis. It assumes that participants have no previous knowledge of statistics or that they have not used it for a long time. | | |
:: Topics Covered
- Introduction to the Analysis of Measurement Uncertainty
- Definition of Uncertainty
- Why is it Important to Measure Uncertainty?
- Characteristics of a Measurement System
- Positioning Uncertainty Analysis in Good Laboratory Practices (GLP)
- Two Approaches to Uncertainty Analysis: Type A and Type B Evaluation
- Statistical Tools for Uncertainty Analysis
- Descriptive Statistics
- Inferential Statistics
- Pertinent Statistical Distributions and their Properties
- The Process of Measuring Uncertainty
- Definition of the « Measurand »
- Identifying the Sources of Uncertainty
- Statistical Approach to Measuring Uncertainty: Type A
- Analytical Approach to Measuring Uncertainty: Type B
- Combining the Two Types of Uncertainty (Combined Uncertainty)
- Interpretating and Communicating the Results
- Related Methods (an Overview)
:: Course ContentAlthough some numerical measures can be determined exactly – like the number of brothers you have, for instance – measures taken in an industrial context, as in quality control, production or R&D, are often subject to a number of different sources of variability. Varying results can therefore be obtained when repeatedly measuring a same sample in a study or experiment.
In addition, more the measurement system is complex (more steps and manipulations required), more the repeated measurements are likely to vary.
For these reasons, it is therefore necessary to calculate a measure of uncertainty associated with the measurements obtained in order to assess the quality of the data. The latter is accomplished by estimating the confidence level associated with a sample value. The data can then be compared to similar data or to a theoretical value. Moreover, the risks or errors associated with the collected data can be minimized so as to avoid incorrect decisions from being made.
Measuring uncertainty is also necessary to ensure the measurement system adheres to the regulatory norms.
This training course covers the techniques for estimating the error associated with a measurement caused by the uncertainties or errors committed in manipulating or in measuring the sample of interest.
The session begins by introducing the types of error which occur when taking a measurement and the criteria for data quality. Statistical tools for assessing and analyzing the variability in the system are presented, namely descriptive and inferential statistics.
The different sources of uncertainty are then discussed to position the analysis of measurement uncertainty among Good Laboratory Practices (GLP). The different types of uncertainty are reviewed and explained in detail: Type A, Type B, combined uncertainty. Finally, how to interpret and report the results based on the latest norms is explained using examples and case studies. Other methods for assessing the quality of the data are also given.
| |
