Adding (Shewhart) control chart test numbers to column in main data table

[#1]

Hello JMP community,

I generate a control chart (Analyse > Quality and Processes > Control Chart Builder) and run all tests (Right click on chart > Warnings > Tests > All Tests) to get the chart below:

I want to add the test numbers that appear next to some of the points, in a new column in the same data table from which the chart was generated, such that each number appears in the same row as the data point it belongs to (data points that have no number in the chart can simply have an empty entry or N/A in this new column). Ultimately, I would like to use this column as a 'colour by' in a 3D plot.

Thank you in advance.

Ahmed 

[#2 · Accepted solution]

Hi Ahmed, 

To get what you want may require a few clicks, but it's not too painful. If you select "Save Summaries" from the red triangle menu for the Control Chart Builder report a new table is generated which should contain all the information you need to join the Test Failures column to the original data table. If this is an Individual Chart, joining the new column could be done with copy/paste since the data should be presented in the same order in both tables. In all cases, Tables->Join could be used to combine the desired column(s). As always, this could be scripted if you need to do this frequently. Hope this helps.

-Jerry

[#3]

Thank you for the prompt reply Jerry!

[#4]

Aside from the technical method to do what you want...Could you help me understand why?  Not trying to be antagonistic, just enhance my knowledge.  I understand the assignable causes for a trend may be different than a point OOC, but the charts are just meant to help identify those, you still need process knowledge to determine what the causes are in any case.

AFAIK, Shewhart did not propose differentiating the different tests.  These were established by a group at the Western Electric Company in 1956. They were intended to provide guidance and perhaps enhance our ability to identify unusual patterns in the data presented as a time series (over and above the control limits interpretation). I'm curious as to why you would differentiate what "rule" was used to identify unusual patterns as evidence of assignable cause variation?  Also the rules often do not apply to 1 point.  For example, on a MR chart an OOC condition is a function of 2 data points.  A trend is a function of 7 data points...etc.

[#5]

Hi statman,

I do not yet have an in-depth understanding of this particular topic, and it seem that you do, so my answer might not quench your curiosity. I work with high-density data points collected from physical 3D objects, and my motivation was simply to see if there are any spacial trends (i.e. a correlation between the locations of the data points on the objects and the rules flagged).

Sorry if this does not answer your question, but I don't know enough yet to fully engage with your question. Thank you for sharing your thoughts though - I'm sure there is something in them for me to learn.

[#6]

@Mark_Bailey and @statman, thank you very much for taking the time to generously elucidate ideas for my guidance. I'll take heed. Much appreciated.

[#7]

A control chart is a tool to generate signals that a process is unstable. You already know that it is not stable. The point of a control chart is not monitoring but to initiate an investigation and implement a correction upon seeing a signal. These charts are also the basis for other kinds of analysis, but I am not sure if it is the best for what you want to know.

There are better tools for finding and assessing associations. They exist for both quantitative and qualitative variables. Many of them fall under the umbrella of 'prediction' and 'classification.' The unsupervised learning models will help you if there is no pre-determined response (e.g., pass, fail). Hierarchical clustering is a good example of a tool that might be useful for your purpose.

Please see the Help > JMP Documentation Library > Predictive Modeling guide for a full explanation and worked examples.

[#8]

A different interpretation of the charts (actually Shewhart's original interpretation).  There are 2 charts that Shewhart created.  The Range chart and the X-bar chart.  These are used together.  The range chart assesses the consistency/stability of the within subgroup variation (which is a function of the variables (x's) changing at within subgroup sampling frequency).  This is done because the within subgroup variation is the basis for comparison.  If the within subgroup variation were inconsistent, then the comparisons would be useless (they would depend on what the within subgroup variables were doing at the time). The X-bar chart is a comparison chart.  It does not assess consistency, but instead compares the within subgroup variation (these are the control limits) to the between subgroup variation, the plotted averages (which is a function of the variables changing at the between subgroup sampling frequency).  If no points are out-of-control, then the within subgroup variation dominates.  This set of x's is where the leverage is.  If there are points out-of-control (or other non-random patterns) then the between subgroup variation dominates.  In this way, control charts helps point investigators towards the subset of x's that have the greatest leverage.  Further sampling or experimentation is necessary to further desegregate the subset of variables to identify the specific causal factors.