topic How to display Cpk for Non-normal Distribution Data in Discussions

How to display Cpk for Non-normal Distribution Data

SpecificityCrow — Thu, 30 Nov 2023 03:34:55 GMT

Hi,

From this tutorial: https://community.jmp.com/t5/Statistical-Thinking-for/Estimating-Capability-for-Nonnormal-Data/ta-p/271889 , JMP will report Ppk only from the process capability platform. However, it was mentioned that we can opt to display Cpk as well by going to the JMP Pro 17 platform settings. I was not able to find the way to adjust this, can anyone help?

Thank you.

Re: How to display Cpk for Non-normal Distribution Data

WebDesignesCrow — Thu, 30 Nov 2023 04:26:16 GMT

For JMP 17 to display Cpk, you need to fit the data to normal distribution.

So, you can click "Continuous fit" -> "fit normal" distribution " -> click the red triangle to perform capability analysis -> specify specs.

Re: How to display Cpk for Non-normal Distribution Data

SpecificityCrow — Fri, 01 Dec 2023 13:10:56 GMT

My dataset are non-normally distributed (has only upper spec limit) and JMP suggested to fit to lognormal distribution. Should I go about it the same way?

Re: How to display Cpk for Non-normal Distribution Data

WebDesignesCrow — Mon, 04 Dec 2023 01:43:55 GMT

Alternatively, you can also force normal distribution by clicking "Process Capability" from the red triangle.

Define USL. JMP will show you the Cpk.

I think only normal distribution will display Cpk.

Cpk = min (Cpu, Cpl).

So, if you have only upper spec limit, Cpk = Cpu because Cpl not exist.

For Non normal distribution (including log normal fitting), JMP will display Ppk.

I prefer to use Ppk because i think it is a better metric compared to Cpk.