Difference between two formulas of process sigma
Six Sigma – iSixSigma › Forums › Old Forums › General › Difference between two formulas of process sigma
 This topic has 2 replies, 3 voices, and was last updated 18 years, 11 months ago by Satya Gopal Kalluri.

AuthorPosts

December 15, 2002 at 8:02 pm #31027
Logan LuoParticipant@LoganLuo Include @LoganLuo in your post and this person will
be notified via email.Does anybody test the difference between following formulas of process sigma?
process Sigma = NORMSINV(1((Total Defects) / (Total Opportunities))) + 1.5
process Sigma = 0.8406 + SQRT(29.37 – 2.221 * (ln(DPMO)))
The second one can not handle DPMO which is greater than 560000.
Can anyone explain the calculation behind NORMSINV(P)?
Thanks
Logan0December 16, 2002 at 3:19 am #81519
Dr. Steve W.Participant@Dr.SteveW. Include @Dr.SteveW. in your post and this person will
be notified via email.it is the inverse funciton of cumulative normal CDF: it maps the (0, 1)
interval into (inf, +inf) and maps .5 to 0, .68 to 1, etc…0December 16, 2002 at 11:48 am #81530
Satya Gopal KalluriMember@SatyaGopalKalluri Include @SatyaGopalKalluri in your post and this person will
be notified via email.Both the formulae attempt to do the same thing. They calculate that point on the xaxis of a standard normal distribution*, to the right of which the area under the said standard normal distribution = (Total Defects) / (Total Opportunities).
The value 1.5 takes into account the long term mean shift of the process.
* Standard Normal Distribution = Normal Distribution with Mean = 0 and Standard Deviation = 1The first formula is calculated by a method used by MS Excel, the details of which I do not know.
I also do not know how the second formula is derived. Maybe it comes from a method that approximates the normal distribution. In fact, I would like to know further details.
But, as far as the second formula is concerned, it does not work above a DPMO value of 560000 because of the error values generated from the SQRT() function.
Error values from the SQRT() function would be generated when the result from the calculation:
[29.37 – 2.221 x ln(DPMO)] gives a negative value. This then depends on the result of ln(DPMO) or the figure of DPMO.
The DPMO figure for which the above difference turns negative is not 560000, but it is:
= EXP(29.37/2.221) = 553364.9868568The difference between the results given by the two formulae is <= 0.01 when 0.9141494 <= DPMO <= 271009
Regards,
Satya Gopal0 
AuthorPosts
The forum ‘General’ is closed to new topics and replies.