How to make UCL appear in Histogram?

[#1]

Hi experts,

I was demonstrating the JMP functionality to my new colleague.

And suddenly hit with below scenario;

If I define the Spec Limit = the line will appear in histogram automatically.

But why, when I define the Control Limit, the line doesn't appear automatically?

I really don't like to go through the scripting details for basic information like control limits to my new colleague.

Is there any preference that I can define rather than insert new reference line manually?

I'm using JMP 17.0

 

Thanks in advance!

[#2 · Accepted solution]

As far as I know, only Spec Limits and Response Limits have the option to automatically added to graphic output.

[#3]

Control limits should only be used in relation to some rational series such as time.  There is no series in a histogram so there should be no control limits.

[#4 · Accepted solution]

Temporal aspects are clearly lost when the temporal structure is neglected with a box plot.

But why is it not possible to generate a plot like this directly via the Distributions platform?

dt = Open( "$SAMPLE_DATA/Quality Control/Phase Historical Data.jmp" );
dt:Site << Data Type( Character );
myControlChart = dt << run  script("Control Chart Builder");
myLimits = myControlChart <<Save Limits("in New Table");
dtsplit = myLimits  << Split(
	Split By( :_LimitsKey ),
	Split( :Force ),
	Group( :Site )
);
dt << Update(
	With( dtsplit ),
	Match Columns( :Site = :Site ),
	Add Columns from Update Table( :_LCL, :_UCL ),
	Replace Columns in Main Table( None )
);
Graph Builder(
	Variables(
		X( :Site ),
		Y( :Force ),
		Y( :_LCL, Position( 1 ) ),
		Y( :_UCL, Position( 1 ) )
	),
	Elements(
		Histogram( X, Y( 1 ), Legend( 14 ) ),
		Line Of Fit( X, Y( 2 ), Y( 3 ), Legend( 15 ) )
	)
)

[#5]

Control chart method, as invented by Shewhart, perform 2 functions:

• Assessing stability of the within subgroup sources/components of variation (this can only be done in a time series R, MR charts), and
• Comparing the within sources of variation to the between (X-bar charts).  The control limits on the X-bar chart are a function of the within subgroup sources (A2*R-bar).

Putting control limits on distributions completely misses the use of the control chart method and should not be done.  Of course, you can put any limits you want on the chart, they just aren't control limits.

[#6]

I like this visualization. Thanks @hogi 

You know, common issues that I've seen in my organization.

They follow statistics to calculate Control Limit & propose control limits that is outside of spec limit (customer requirement) which I think is ridiculous.

So, I just want to visualize the comparison (easy to explain)

[#7]

Yes, to detect analysis errors and discuss them, the visualization in Jmp is a great advantage!

How good that we have GraphBuilder :)
It can replace many of the other [sometimes limited] analysis platforms :)