cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Check out the JMP® Marketplace featured Capability Explorer add-in
Choose Language Hide Translation Bar
chris_G_ttu
Level II

How to pick the correct distribution for a capability index plot

Hello,

 

How do I pick the correct distribution for a capability index plot? For example, How do I know to use a normal, lognormal or weibull distribution. How would the type of distribution affect the ppk?

2 ACCEPTED SOLUTIONS

Accepted Solutions
Peter_Bartell
Level VIII

Re: How to pick the correct distrabution for a capability index plot

Another potential path for you to consider in addition to @txnelson's idea is to use (if you have JMP version 13) the Process Capability platform sub platforms around fitting distributions and non normal process capability indices. Here is the link to the JMP online documentation for this area of analysis capabilities:

 

http://www.jmp.com/support/help/13-2/Distribution_Options.shtml#461340

 

View solution in original post

Kevin_Anderson
Level VI

Re: How to pick the correct distrabution for a capability index plot

Hi, chris_G_ttu!

 

Jim is correct (like always! ;-) about JMP determining the distribution type for you.  But you should also evaluate the theory and context of the data generation mechanisms of your process and allow that evaluation to inform your choice.  Don't blindly follow the recommendations of any software, even one as righteous as JMP.

 

I recommend you consult a statistician.  There are many references (Montgomery, Wheeler, Kotz & Lovelace) regarding capability metrics and Cpk/Ppk that detail the effects of different distributions.  It is a broad and deep subject that is difficult to cover adequately in a discussion forum.

View solution in original post

5 REPLIES 5
txnelson
Super User

Re: How to pick the correct distrabution for a capability index plot

The easiest way to determine the destribution you have, is to go to the Distribution Platform, select the columns you want to be analyzed, and then once the disttribution histograms are displayed, go to the red triangle and select

     Continuous Distribution==>All

The platform will then attempt to determine what distribution your data are.

Jim
chris_G_ttu
Level II

Re: How to pick the correct distrabution for a capability index plot

Thanks for the response, I have alredy found the best distrabution for my veriables. I am wondering how the dsitrabution affects Ppk and which distrabution to pick to get the correct Ppk. Is the correct distrabution in a capbility index plot always the distrbution that fits the best? 

Peter_Bartell
Level VIII

Re: How to pick the correct distrabution for a capability index plot

Another potential path for you to consider in addition to @txnelson's idea is to use (if you have JMP version 13) the Process Capability platform sub platforms around fitting distributions and non normal process capability indices. Here is the link to the JMP online documentation for this area of analysis capabilities:

 

http://www.jmp.com/support/help/13-2/Distribution_Options.shtml#461340

 

Kevin_Anderson
Level VI

Re: How to pick the correct distrabution for a capability index plot

Hi, chris_G_ttu!

 

Jim is correct (like always! ;-) about JMP determining the distribution type for you.  But you should also evaluate the theory and context of the data generation mechanisms of your process and allow that evaluation to inform your choice.  Don't blindly follow the recommendations of any software, even one as righteous as JMP.

 

I recommend you consult a statistician.  There are many references (Montgomery, Wheeler, Kotz & Lovelace) regarding capability metrics and Cpk/Ppk that detail the effects of different distributions.  It is a broad and deep subject that is difficult to cover adequately in a discussion forum.

Peter_Bartell
Level VIII

Re: How to pick the correct distrabution for a capability index plot

To pile onto @Kevin_Andersonn's input...ALWAYS use the most effective 'goodness of fit' tool ever invented...your eyes.