Hi all,
I've had a couple of questions in from a customer who is getting significantly different values for Ppk than is being obtained in their verification work outside JMP. The difference between these calculated Ppk values are generally greater than 0.2.
So that they can try and validate their work, the 2 questions I've had for their nonparametric system are:
- Which method (Scott, Silverman etc.) is being used to select the nonparametric density bandwidth?
- Once the kernel density estimate has been generated, what formula is being used on it to generate Ppk etc.?
Having been through the documentation I could find, I've been unable to answer either question, so thought I'd ask the collective minds on the community for help!
Thanks in advance,
Stuart
Hi Stuart,
For question 2, this doc page has the details: https://www.jmp.com/support/help/en/16.2/index.shtml#page/jmp/capability-indices-for-nonnormal-distributions-percentile-and-zscore-methods.shtml#
Question 1 does not seem to be answered by the documentation, I agree. I would submit this to tech support with customer on cc.
Jordan
