Re: quality control
- From: "Old Mac User" <chendrixstats@xxxxxxxxx>
- Date: 14 Nov 2006 11:17:12 -0800
Question: How many pieces must be sampled/inspected... with no defects
in those pieces... to be 99% confident that the process has a quality
level of 0.999 (meaning... the true underlying defect rate is 1,000 ppm
or less? 1,000 ppm is equivalent to 1 in 1,000 or 0.1% defective)
The following equation can be derived. If you want the derivation,
please contact me by e-mail.
n = [Ln(1 - c)]/[Ln(q)]
where
c = confidence level
q = quality level
n = sample size with no defects in the entire sample
n = [Ln(1 - 0.99)]/[Ln(0.999)]
n = -0.460517/-0.001
n = 4,605
Oops!! That's larger than the population that's to be sampled.
And yes, I know you mentioned 0.01%, not 0.1%.
The equation n = [Ln(1 - c)]/[Ln(q)] is based on the assumption that
the sample is a modest fraction of the total population. Say, the
sample is less than 15% of the population.
OMU
Old Mac User wrote:
I see there are several posts concerning this. There is a specific
equation that applies here, and we can use it to calculate the
appropriate sample size. No muss, no fuss, no rain dances about
"priors", etc.
Sadly, as several have noted, the answer is going to be unpleasant.
You'll need a very large sample size.
I'll post that equation and definition of terms later today. OMU
Frank wrote:
If I know a product fails .01% of the time and I have 1500 items I'm
running through a process. How many items do I need to check with,
say, 99% confidence that all the items are built correctly.
I'm thinking either Poison or Geometic distribution is applicable, but
what would the formula be. Also, what key words should I be searching
for. I know I've come across this problem many years ago but stuck on
finding where.
Thanks,
F
.
- Follow-Ups:
- Re: quality control
- From: David Winsemius
- Re: quality control
- References:
- quality control
- From: Frank
- Re: quality control
- From: Old Mac User
- quality control
- Prev by Date: Re: quality control
- Next by Date: Finding Significant Data Points in a Time Series
- Previous by thread: Re: quality control
- Next by thread: Re: quality control
- Index(es):
Relevant Pages
|