Visualize Your Process Behaviour: Making Sense of Data Chapter 7
In this chapter, the author introduces two types of variation, predictable and unpredictable, and the improvement strategies applied to each variation respectively. XmR chart is also presented as a tool to distinguish the types of variation. However, the author does not explain why XmR chart works in this Chapter. This made me a bit confused and not being persuaded to use the chart. Then, I look at the content page again and find that in Chapter 11, the author will talk about What Makes the XmR Chart Work?. So, I decided to read Chapter 11 directly after finishing Chapter 7. I need to understand the rationale behind it.
| Predictable Process | Unpredictable Process | |
| Reasons | Process subjects to many cause-and-effect relationships where no one cause is dominant over the others. | Process subjects to some dominant causes which are called assignable causes. |
| Improvements needed | Improvement will only come by changing a major portion of the process. It will be a waste of time to look for assignable causes (a particular cause). | Improvement will come by finding and removing the assignable causes. Changing a major portion of the process will be premature. |
But how can we know a process is predictable or unpredictable. The author then introduces a tool called XmR chart. The basic concept of XmR chart is to compute an upper bound and lower bound of a time series. Anything outside the bounds is unpredictable.
The red line in the top chart is the original time series. The steps to reproduce the charts are as follow:
- Compute the Moving Ranges (mR) by calculating the running difference between two points.
- Plot the mR in the red line of the bottom chart
- Compute the average of mR. Plot it as the blue line.
- Compute the Upper Range Limit (URL) = 3.27 * the average of mR. Plot it as the green line.
- Compute the Upper Natural Process Limit (UNPL) = average of the time series + (2.66 * average of mR). Plot it as the green line in the top chart.
- Compute the Lower Natureal Process Limit (LNPL) = average of the time series – (2.66 * average of mR). Plot it as the orange line in the top chart.
In this chapter, the author does not explain the reason why XmR Chart work. I will jump a few chapters and read Chapter 11: What Makes the XmR Chart Work?.
Comments ()