I saw this on a six-sigma Process Capability training that i attended.
Cpk for an attribute data was totally news for me.
Here is thier way of doing it:
Step1: From attribute data, compute process yield.
Step2: From the yield information, determine Z-value correspoding to the same
area under the Normal Curve. Call this Sigma Level. In excel SigmaLevel=NORMINV(YIELD,0,1)
Step3: Remember that Z-score = (X-Mean)/(Sigma). This is almost similar to Cpu=(USL-Mean)/(3*Sigma) except for the factor 1/3. Therefore it is possible to think that Cpk = Sigma Level/3.
What do you think? Is that logic valid? Is this method generally acceptable?
Cpk for an attribute data was totally news for me.
Here is thier way of doing it:
Step1: From attribute data, compute process yield.
Step2: From the yield information, determine Z-value correspoding to the same
area under the Normal Curve. Call this Sigma Level. In excel SigmaLevel=NORMINV(YIELD,0,1)
Step3: Remember that Z-score = (X-Mean)/(Sigma). This is almost similar to Cpu=(USL-Mean)/(3*Sigma) except for the factor 1/3. Therefore it is possible to think that Cpk = Sigma Level/3.
What do you think? Is that logic valid? Is this method generally acceptable?