Skip to main content
ARS Home » Research » Publications at this Location » Publication #123370
Title: THE USE OF CONTROL CHARTS TO MONITOR REGRESSIONS PARAMETERS: TWO PRACTICALAPPLICATIONS

Author
item STEWARDSON, S - UNIVERSITY OF NEWCASTLE
item DREWETT, L - CORUS GROUP
item SILVA, L - INSTITUTO DE ENGENHARIA
item MERTENS, J - THYSSEN KRUPP
item Malone, Robert - Rob
item Bonta, James - Jim

Submitted to: Meeting Abstract
Publication Type: Proceedings
Publication Acceptance Date: 9/10/2000
Publication Date: N/A
Citation: N/A

Interpretive Summary:

Technical Abstract: There are many business and industrial situations in which some linear regression curve is used to describe the nature of a process. Typical examples might include; a standard curve used to link some measured response to a true value, a measure of performance improvement over time, or a way of assessing some material property under cumulative stress. In each of these there is often a need to establish if changes have occurred over time to the relationship between the predictor variable and the response. A change may occur because of problems with the measurement process itself, changes in the effect that the predictor is having due to material changes, or a dropping off in improvement rates over time. An effective way of monitoring data to find these changes is by the use of control charts modified to handle the parameters of a statistical model rather than the raw data upon which it is based. The paper demonstrates two unrelated examples of this method, showing how the model parameters were discovered, the setting of control limits for these and the use of the charts in practice. Each application has lead to a significant improvement in the monitoring of the processes involved, the control of the objectives of each work situation and better understanding of the underlying nature of each process. In both cases improvements to international standards were suggested or noticed. The cases show how combining relatively simple statistical concepts together can be used to significantly improve peoples understanding and control of key processes and give greater confidence in the tests or measurement being taken.