A single X chart outperforming the joint X & R and X & S charts for monitoring mean and variance
Salah Haridy, Yanjing Ou, Zhang Wu, Michael B. C. Khoo
Abstract
Salah Haridy, Yanjing Ou, Zhang Wu, Michael B. C. Khoo
Abstract
The Shewhart X & R and X & S control charts have traditionally been used for detecting mean shift δμ and standard deviation shift δσ. This article studies and compares the overall performance of the X chart with that of the X & R and X & S charts, as well as the X&MR chart. The comparative study led to surprising results that contradict the conventional wisdom in Statistical Process Control (SPC) niche. It is found that the simplest single X chart (i.e., the X chart with a sample size n = 1) is always the optimal version of the X chart for detecting δμ and δσ. Moreover, the single X chart even outperforms the joint X & R and X & S charts in overall detection effectiveness. On average, the X chart is more effective than the X & R and X & S charts by around 5% under different circumstances. Most importantly, the X chart is very simple to understand, implement and design. As a result, it may be highly preferred for many SPC applications, in which both the mean and variance of a variable need to be monitored.
OpenAlex reports 10 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
The Shewhart X & R and X & S control charts have traditionally been used for detecting mean shift δμ and standard deviation shift δσ. This article studies and compares the overall performance of the X chart with that of the X & R and X & S charts, as well as the X&MR chart. The comparative study led to surprising results that contradict the conventional wisdom in Statistical Process Control (SPC) niche. It is found that the simplest single X chart (i.e., the X chart with a sample size n = 1) is always the optimal version of the X chart for detecting δμ and δσ. Moreover, the single X chart even outperforms the joint X & R and X & S charts in overall detection effectiveness. On average, the X chart is more effective than the X & R and X & S charts by around 5% under different circumstances. Most importantly, the X chart is very simple to understand, implement and design. As a result, it may be highly preferred for many SPC applications, in which both the mean and variance of a variable need to be monitored.
Key concepts: Chart, X-bar chart, \bar x and R chart, Control chart, Computer science, Standard deviation, Variance (accounting), Algorithm