Abstract:
In quality control, the aim of periodical sampling of products from processes is to verify that the
products are of high quality. In this paper, we study the performances of control charts when the
probability of observing a defective product follows a Zero-Inflated Poisson (ZIP) distribution, i.e.,
a distribution that is similar to a Poisson distribution but with an excess number of zeros. We
compare the performances of a c-Chart with Jeffreys prior interval (cJ -Chart), a c-Chart with ZIP
model (cZIP-Chart) and a Shewhart c-Chart (c-Chart) for processes with varying proportions of
observed excess zeros, namely,ω = 0.3, 0.4, 0.5 and 0.6. We looked at processes in which there was
no mean shift in amount of defectives and in which there was a mean shift, and for each case we
evaluated the efficiencies of the control charts using the average run length (ARL) and the average
of coverage probability (ACP). The results showed that when there was no mean shift, then the cJ
Chart gave good efficiency, with efficiency depending onω . When there was a shift, then the c-
Chart was the most effective for detecting the shift, but the ACP of the c-Chart was found to be the
least effective in estimating process parameters.