CUSCORE Charts and Their Application to Monitoring a Sawing Process
ABSTRACT:
CUSCORE charts were introduced by Box and Ramirez (1992) as tools
directed to detecting the presence of specific disturbances
feared in advance in a process. Thus different charts could be
used in a given application depending on how many different problems
one wishes to keep a close eye on. Industrial applications were the
first ones in mind but the methods apply to any other processes, e.g.
services, administration, etc. From a methodological perspective,
the charts can be seen as the cumulative sums of Fisher's score
statistics. In this talk, we discuss the ideas behind the charts,
their inner workings and their optimality under a plausible
criterion. We illustrate their use in monitoring a sawing process
from the lumber industry where one is interested in detecting
specific sawing defects such as chipper marks, roll marks, wane and
bandsaw marks. A dataset provided by the Faculty of Forestry at
UBC will be used for illustration.
Part of the work was done jointly with Alberto Ferrer and Yongmin Yu.
About the Speaker
Roman Viveros-Aguilera is a professor of statistics in the
Department of Mathematics and Statistics at McMaster University.
He obtained a bachelor in mathematics from Universidad
Veracruzana (Mexico), a master in mathematics from the
Center for research and Advances Studies (Mexico) and
a PhD in statistics from University of Waterloo. He is
interested in statististical inference, industrial statistics
and applications. Currently, he is the editor of Liason, the
quarterly information bulletin of the Statistical Society of Canada.
References
Box, G.P. & Luceno, A. (1997). Statistical Control by Monitoring
Monitoring and Feedback Adjustment, Wiley: New York.
Box, G.P. & Ramirez, J. (1992). Cumulative Score Charts.
Quality and Reliability Engineering International 8, pp. 17-27.
Viveros, R. & Ferrer, A. (2002) Intrinsic Sensitivity and Optimal
seeds for Constructing Efficient Control Charts. Manuscript available
from first author.