topic Re: cpK/ppk in Discussions

cpK/ppk

Pops — Tue, 23 May 2017 12:21:03 GMT

Is it correct to calculate cpk when I have no subgroup in the process data (grouping=1), meaning the sampling is continous, like tablet weight taken one after the other in one lot? How is cpk calculated in this case (the withingroup variation)?

Re: cpK/ppk

Peter_Bartell — Fri, 19 Jan 2018 19:17:42 GMT

Here's a reference for you in the JMP Online Documentation.

http://www.jmp.com/support/help/13/Capability_Analysis_2.shtml#1161749

As of January 2018, use this link:

http://www.jmp.com/support/help/13-2/Capability_Analysis_2.shtml#1161749

Re: cpK/ppk

Peter_Bartell — Tue, 23 May 2017 20:00:48 GMT

Correct? That's a different question. The 'correctness' of an approach is dependent on many aspects to the process capability analysis problem. For example: 1. Does your process data exhibit any evidence of assignable cause variation? 2. Does your data appear to be normally distributed? The answer to these questions influences the approach that is optimal for calculating process capability indices over and above the sheer mathematics associated with the reference in my first reply.

Re: cpK/ppk

Pops — Wed, 24 May 2017 09:59:22 GMT

Thank You

Data are normal distributed with no assignable cause variation. I assume the answer I was looking for (found in Process Capability in JMP 2012) is that if I do not group data JMP will automatically use moving range subgroup size 2 to calculate cpk.