Cpk additive?

Hello John,

I apologise. I did not see your response to my initial reply! In the interim, I think that John A. has effectively answered your question. If you need to see the capability of each machine/shift/material batch etc. then as you say, you have already calculated this via Minitab.

If for whatever reason you wish to see the aggregration, then crunch all the data, and you will get an overall capability measure.

I would think that Isaac Newton (really) at Minitab in Coventry can help you if you have further problems with the calcs.

Do what Paul A. says - group all of the data together, for continuous data (or variable as it is often called) plough it into Minitab ([url]www.minitab.co.uk[/url])and run the Capability Analysis for the respective type of distribution. Otherwise, as indicated by other 1SS.com users, count the number of out of specification datapoints and express as a % or DPMO attribute data figure, this will then easily convert to a Z value for capability. Is your data long or short term? if it is short term don't forget to take 1.5 off the Z value to make it into a Zlt (long term) figure.

If you want an overall cpk just lump all the data together when you calc the cpk figure. After all this is the result that the customer is seeing.

Merry Christmas Liz,
Thanks for replying,Thats a good idea on the dpmo. I'll give it a go.

[quote:bf42550e2b="Liz Ferguson (MBB)"]You cannot average Capability Scores, they are not a linear measure and as such should never be averaged. You could look at averaging the defect rate or dpmo which could then be converted to a sigma score (or Capability Score).[/quote:bf42550e2b]

Many Happy Returns Roy,
You are correct and saying I have Cpk values for each machine with Minitab ([url]www.minitab.co.uk[/url]). It is to get a overall Cpk value I'm trying to get too..If you could advise on that.

Merry Xmas

[quote:465ccd5d0e="Anonymous"]Seasons Greetings John,

Averaging out the capability from each cell or machine would be extremely dangerous as you are sure to mask some root causes of variation. The danger also exists that even if you do see variation, you will treat it incorrectly, and probably end up in making the overall worse off than before.

I would start by running a control chart on each of the process variables per machine/plant/cell or whatever. If you have already done that (you don't say what types of data you have), you can then calculate the Cpk for each cell/machine/etc and the overall. Minitab or similar will run this calculation as well as the analysis of variation thereafter.

Roy[/quote:465ccd5d0e]

Seasons Greetings John,

Averaging out the capability from each cell or machine would be extremely dangerous as you are sure to mask some root causes of variation. The danger also exists that even if you do see variation, you will treat it incorrectly, and probably end up in making the overall worse off than before.

I would start by running a control chart on each of the process variables per machine/plant/cell or whatever. If you have already done that (you don't say what types of data you have), you can then calculate the Cpk for each cell/machine/etc and the overall. Minitab ([url]www.minitab.co.uk[/url]) or similar will run this calculation as well as the analysis of variation thereafter.

Roy

You cannot average Capability Scores, they are not a linear measure and as such should never be averaged. You could look at averaging the defect rate or dpmo which could then be converted to a sigma score (or Capability Score).