topic Re: Cpk & Ppk in Discussions

Cpk & Ppk

RegressionBear1 — Wed, 22 Jan 2025 06:28:06 GMT

I want to calculate the Cpk value for some data，but sometimes it only shows the Ppk value.

As shown in the figure below, there are two sets of data of the same type with the same specification limits. Through smooth curve fitting and nonparametric density capability analysis, one shows Cpk while the other shows Ppk.

Re: Cpk & Ppk

Ben_BarrIngh — Wed, 22 Jan 2025 09:34:23 GMT

@RegressionBear1 , would it be possible to provide the dataset to look at? From a cursory glance, the reason you may only be seeing Ppk is because you have non-normal data, Cpk works on the assumption that the data is normal, so the control limits are equally spaced (see the diagram below - if the limits in blue are not equally spaced around the centre, Cpk cant be calculated).

There are some useful discussions here:
https://community.jmp.com/t5/Discussions/Why-JMP-Cpk-Ppk-is-not-consistent-with-non-conformance/td-p/679860

https://community.jmp.com/t5/Discussions/How-to-display-Cpk-for-Non-normal-Distribution-Data/td-p/704169

https://www.jmp.com/support/help/en/17.2/#page/jmp/individual-detail-reports.shtml#

Thanks,

Ben

Re: Cpk & Ppk

RegressionBear1 — Thu, 23 Jan 2025 03:11:12 GMT

Thank you for your reply. I have attached the data. What puzzles me is that the "moisture" data in both tables are similar and do not follow a normal distribution. After fitting a smooth curve and setting the same specification limits, one dataset yields a Cpk value, while the other can only provide a Ppk value.

Re: Cpk & Ppk

Ben_BarrIngh — Thu, 23 Jan 2025 11:40:27 GMT

Hi @RegressionBear1 ,

From what I can tell of the data, I've had a look through the data - the parts that you are using are from v15 of JMP, which are now 'Legacy Fitters', but I've tried to redo your work in v18.0.

I tried to recreate your steps and did find that the term 'PPK' turns up in the analysis instead of CPK - I suspect this may be due to you using a script to create the analyses. I would suggest you do two things:

a) Update to v18 to take advantage of the improvements to fits

b) Contact support@jmp.com to highlight the problems with the data.

Many thanks,

Ben