Process Capability Spec Limit Setting

[#1]

Hi, 

I have the attached JMP sheet that I'm trying to run process capability analysis for. The process has different specification limits for each "condition" but when I run the analysis it only allowed me to set one set of limits. Is there any way to set different limit for different conditions? Or I will have to break into multiple columns as "Performance@20", "Performance@40", etc..?

Thanks,

Gary

[#2]

Gary,

There may be a better way to do this, but I believe you will have "Split" your data and then enter the spec limits separately.  The good thing though is if you are using JMP 13 or 14 you can use Process Screening under Analyze > Screening to run all of the capability analysis at once.  If you have non-normal data you will get a more reliable answer if you run your Capability Analysis under Analyze > Quality and Process > Capability Analysis.  I did a quick split of your data and analyzed the distributions for the 5 performance conditions.  Only the "80" condition appears to be normally distributed.

There is some good documentation in Help > Books > Predictive and Specialized Modeling > Process Screening.

HTH

[#3]

Bill, 

Thanks for getting back to me. Two follow up questions:

1. Was there an easy way to "split" the data into multiple columns in JMP?

2. When I ran distribution on the dataset and looked at the quantile plot they all look normally distributed except the "20". Am I reading this incorrectly? Any other metrics to better determine whether it's normally distributed?

Thanks,
Gary

[#4]

Gary,

For the splitting, use Tables > Split. Put Condition in Split By and Perfomance in Split Columns.

From there go to Analyze > Distribution and select the 5 Condition columns.  Click OK

Hold down the Control Key (Windows) and then click the red hot spot by "20".  Go to Continuous Fit near the bottom and toggle over to "All".  This will find the best overall distribution for the 5 columns.  I would suggest you do regular and non-normal capability analysis the Normal Mixture distributions and see if they come out different from each other.

Bill

[#5]

You do not need to split your table to accomplish this.  You can use the Process Screening platform directly.  I have written a blog post that details how to accomplish this.

https://community.jmp.com/t5/JMPer-Cable/Process-specification-limits-that-differ-based-on-a-groupin...

[#6]

Thanks Tanya!

[#7]

Tonya, 

I tried what you described on the blog and created a new spec limit jmp file but I'm unable to load this jmp file at the attached screenshot. Let me know if I'm doing anything wrong.

Thanks,

Gary

[#8 · Accepted solution]

Your spec limits table does not have the appropriate format.  In my blog post you will see that the spec limits data table has 5 columns.  Your spec limits table only has 4 columns.  You are missing the Process column.  Since process screening can handle more than 1 process (response), you need a column in the spec limits table to tell process screening which response the data table entry is referring to.  Add a column to your spec limits data table.  You can call it anything, but if you name it Process, that is a keyword for JMP and it will automatically fill that column into the appropriate section of the dialog.  Each entry in the spec limits data table for Process should be Performance.

[#9]

Thanks Tonya,

I tried what you described and it only seemed to take the first row of spec defined. Am I doing something wrong?

[#10]

You are using the Process Capability platform.  The blog post is about the Process Screening platform.  Process Capability does not allow grouping variables.  You will need to use Process Screening if you want to use your grouping variable.