how to calculate sigma for attribute data of process
Six Sigma – iSixSigma › Forums › Old Forums › General › how to calculate sigma for attribute data of process
 This topic has 6 replies, 4 voices, and was last updated 16 years, 11 months ago by Ruddy.

AuthorPosts

October 23, 2005 at 12:44 pm #41151
how to calculate sigma for attribute data of process
I have a process is coating for the production. Totally, I had 10 products, the defect is coating spoliation, Now the point is some product have 0 defect, someone have 3 defects and so on, I used Mintab calculate it, total defects are 17, opportunity is 1, unit is 10, but I can’t get result,defect opportunity uint 1 1 1 3 1 1 1 1 1 1 1 1 0 1 1 2 1 1 1 1 1 5 1 1 0 1 1 3 1 1it show** Error ** Incorrect data – result contains a value of PPM inexcess of 1,000,000 (this is impossible);Execution aborted.I think the problem is “opportunity”? but how can I define the opportunity?, anyway the defect is same kind and have a some standard(only compare with a sample about it’s dimension)Does anybody can give me some suggestion? thanks
0October 23, 2005 at 3:03 pm #128704If the opportunity is one, the it is impossible for a single unit to have three. Isn’t opportunity in this case how many ways a single unit can be defective?Dave
0October 23, 2005 at 3:52 pm #128706
Ken FeldmanParticipant@Darth Include @Darth in your post and this person will
be notified via email.If you are assuming an opportunity of 1 then you are dealing with defectives not defects. It is either good or bad. If you want to do that then just take the number of bad ones regardless of how many defects it had. If you want to use defects, then you have to consider how many different types of defects are possible and that becomes your opportunity.
0October 24, 2005 at 2:56 am #128720Model it as a Poisson distribution with a mean of 1.7.We did the distribution of paint defects/unit area and did a square root transformation to get a normal distribution (I think).Just an idea.Got to go.BTDT
0October 24, 2005 at 11:01 am #128735thanks for your suggestion.
Now the point is we want to reduce the defect sas possible as we can( because it is costly for us). I think it should belong to attribute data. So does anybody can give me a clear way to define opportunity or how to calcualte it.
thanks0October 25, 2005 at 2:47 pm #128819Michael:Sorry for the delay.As I pointed out before, your data can be modeled using a Poisson distribution where the mean number of defects per unit (DPU). When you have a Poisson distribution, you can calculate the probability of finding an item with zero defects using the equation;p(0 defects) = exp(DPU)Using your data, I count 17 total defects in the 10 units of your sample. The DPU is 1.7 and I get p(0 defects) = 0.182684 all other units will have AT LEAST one defect and be unacceptable to the customer.The number of items with AT LEAST one defect is 10.182684 = 0.817316DPMO = 817316This extends Darth’s point about defects and defectives. When a unit has at least one defect, then it is not acceptable to the customer. Counting the number of defects per unit, however, allows you to see if the process is improving at the process level, even though the customer sees the entire unit.When it comes to screening for Vital Xs, use a Chisquare test to see if there are differences in the numbers of defects per unit broken out by subgroups.Does that help? Let me know.Cheers, BTDT
0October 28, 2005 at 9:01 am #129041Hi BTDT,
It is so nice to get your kind help, I do appreciate that.
Michael0 
AuthorPosts
The forum ‘General’ is closed to new topics and replies.