 |
A QC chart typically depicts a random time series fluctuating around a characteristic value. This is because an individual entry is determined by the interplay of many factors, most unknown or undocumented, which usually balance each other but not exactly. There is a probability, however, that the effects of several factors will coincide to raise or lower the entry to a level outside the permitted range. If the entries are independent having no trend or periodicity, the probability of such an event can be estimated from their distribution. This can be demonstrated with industrial records, by comparing the estimated frequencies of these events with those observed in fresh data. The estimation procedure has been automated and posted as freeware on the Internet. Free software is also available for demonstrating that if the factors' effects were multiplicative, the entries' distribution would be approximately lognormal.
|
|
For the newest resources, visit Wolfram Repositories and Archives »
