Process Capability - Interactive Capability Plot

[#1]

Dear JMP Community,

here my need.

I performed: Data Distribution -> Process Capability. I obtained Process Capability histogram & information (Sigma Capability, Nonconformance,...).

Then I selected: Interactive Capability Plot.

If in the Capability Analysis I selected Normal Distribution: no problem... I can have the Interactive Capability Plot.

But if If in the Capability Analysis I selected other distributions (lognormal, weibull,...): I do not find the Interactive Capability Plot option.

Is it correct or I perform some error?

Thanks in advance for your feedback.

Best Regards,

Simone 

[#2]

Hi @Simon_Italy ,

  Welcome to the community!

  A brief answer to your question is that it is correct and you did not make an error. The interactive capability plot option is only available if you've selected a normal distribution.

Hope this helps!,

DS

[#3]

You can mostly get there from here...

Try this:

1. Fit continuous distributions > Fit all

2. Save the distribution formula back to your table

    In the last row of that new table, enter you lower and or upper spec limit, to get the transformed values.
    Save those values into the spec limits property for your new column.

3. Run Distribution on the new distribution formula column, it should be more or less normal now, and then the assumptions of the interactive capability profiler will work... except that it will be in the transformed units.

[#4]

Dear Byron,

thanks for this "work around" (useful).

I will try to use it for my needs.

Best Regards,

Simone

[#5]

Dear DS,

thanks for the welcome and feedback (useful).

Do you know the reason that this feature is available only for Normal distribution and not for others?

Is it linked to a statisitcal/techincal constrain... or only a JMP decision?

Thanks

Have you a good day.

Best regards,

Simone

[#6]

Hi @Simon_Italy ,

  I'm not a JMP employee, so I can't say for sure as to what the reason is for the interaction capability plot only working with the normal distribution, but as Byron alluded to in his reply, "...the assumptions of the interaction profiler will work".

  So, it appears as if there are some assumptions that go into certain calculations and how the interactive profiler is updated. Likely due to the fact that normal distributions have "nicely" defined parameters like standard deviation and mean. Distributions that are very different from normal tend to have much more complicated mathematical definitions for the parameters that define the distribution function. But, this is just my guess as to why.

Hope this helps,

DS