Student Seminar Series - July 18, 2005
University of Minnesota
School of Statistics
College of Liberal Arts

Performance Comparison of EWMA Chart Based on Log and Natural Scaled Sample Variance


Liying Zhang


Monday, July 18, 2005
9:00 AM, 120 Amundson Hall
Minneapolis, East Bank Campus

Refreshments at 8:30 AM
300 Ford Hall


Abstract

 The Exponentially Weighted Moving Average (EWMA) control chart is effective in detecting shifts in the process mean and standard deviation. For monitoring process standard deviation, the EWMA control charts can be constructed by using an EWMA based on the sample variance naturally or the log transformation of the sample variance. In the paper we compare the performance of the two EWMA control charts in terms of the average run length (ARL). Fortran programs for ARL calculation and their outputs are discussed. The examples and comparison graphs show that the EWMA based on natural-scaled sample variance reacts to the shifts in process variability faster than the EWMA based on log-scaled sample variance does, particularly when the shifts are small.