0

The various control char constants are usually given for various population sizes, often up to 25.

I'd like to have more flexibility; is there à way to compute those instead?

Thanks

mdewey
  • 16,541
  • 22
  • 30
  • 57
Michael Hooreman
  • 121
  • 8
  • 2
    Please indicate which "constants" you are referring to. (There are many, depending on the type of control chart.) The answer is yes, these are computable; see the comment at http://stats.stackexchange.com/questions/184034/estimating-process-standard-deviation-for-a-quality-control-chart-what-is-s-c#comment479639_184034 for instance. – whuber Jan 12 '17 at 18:01
  • Basically every constant used for any control chart. I'm creating a general purpose python library for SPC. – Michael Hooreman Feb 03 '17 at 08:09
  • That would comprise an entire book on quality control, making your question overly broad to be answered here. – whuber Feb 03 '17 at 14:28

1 Answers1

1

There are several sources for such calculations. Sadly, they tend to not all be published together.

One source is Manual on Presentation of Data and Control Chart Analysis, Dean V. Neubauer, Editor. Introduction to Statistical Quality Control by D. C. Montgomery has most of the formulae you are looking for.

It may be easiest to check out a handful of books on the subject through your local library and see if you can piece everything together that you need.

On the bright side, you should almost never need a sampling plan that examines samples of more than 25 units at a time.

Tavrock
  • 1,552
  • 8
  • 27
  • Thanks a lot. +1 for the "bright side". In fact, I already have the data points (and there might be a lot), but by reading you, I think I will consider raising a warning if we work with more than 25 points. – Michael Hooreman Feb 03 '17 at 08:10