Shewhart further identified two types of mistakes possible in managing variation: treating a common cause as special and a special cause as a common cause.
Later, W. Edwards Deming estimated that a lack of an understanding of variation resulted in situations where 95% of management actions result in no improvement. Referred to as “tampering,” action taken to compensate for variation within the control limits of a stable system increases, rather than decreases, variation.
A control chart is the visual representation of the SPC and is allot with; the Upper control limit (UCL), the lower control limit (LCL) the ideal target, which is the central measure or limits of the measurement. Data points of the process are ideally supposed to constrain mean and within the boundaries of the UCL and The LCL to meet acceptance criteria.
Every deviation from the central limit is added or subtracted from the standard error from the mean, bringing the result closer to UCL or LCL.
For simplicity, the control chart provides 3* standard error (SE) towards the UCL or LCL based on the confidence level of the sample distribution. The Confidence interval determines the level of risk to which we want to accept errors.
Figure 2
Explanation of the SPC patterns
“It is imperative to note that; the standard error (SE) is to the distribution of many samples means as the standard deviation (sd) is to a distribution of scores in one sample. The SE calculates a confidence interval around a particular sample means. This confidence interval tells us how confident we are that the true population mean ( µ ) falls within a given range. For example, if the results of a survey on general radio listening show average daily listening of “37 minutes, plus-or-minus 4.5 minutes at the 95% confidence level,” we would say that we are 95% certain that the true population means ( µ ) is between 32.5 and 41.5 minutes.”
Figure 3.
Confidence interval associated with standard error
Wikipedia contributors. (2022, August 20). 68–95–99.7 rule. In Wikipedia, the Free Encyclopedia.
Installing the “qcc” library
#install the “qcc” package
Install.packages(‘’qcc’’)
# Upload the “qcc” package
Library (‘’qcc’’)
# using the available in the “qcc” package data called pistonrings
Figure 4.
Pistonrings datasets.
Figure 5.
Calculating the average diameter of a sample of each subgroup
Figure 6.
Transformed the tables in “qcc” friendly format in terms of the diameter of each subgroup
Figure 7.
Code for Xbar S chart and diagnostic output
Figure 8.
The Xbar s chart
Figure 8B.
Using code “qcc (diameter, type=” S”): The given code is used to check the control status of figure 8, and it shows that the process is in control. From the graph, it can be identified that the outlier is due to a random (sporadic) cause, or the error is not common or inherent to the process. This error due is a special cause, as explained in figure 2.
Figure 9.
Code for Xbar R chart and diagnostic output
Figure 10.
Xbar R chart
Figure 10B.
Similarly, using code “qcc (diameter, type=”R”): the process is in control using the R chart; the cause of the error is not inherent but a common cause.
Miscellaneous insight
Figure 11 shows how the data is distributed and can easily show where the anomaly occurred. Figure 12A, 12B, and 12C also show outliers, the minimum the maximum values recorded.
Figure 11.
Data distribution
Figure 12A.
Summary statistics
Figure 12B.
Minimum and maximum
Figure 12C.
Boxplot of subgroup
Conclusion
The cause of the outliers is due to a special cause, and no major modification of the process is required. However, it requires an investigation so that a cause is assigned to the source of the cause.
Awesome!