Sampling plans for the zero-inflated Poisson distribution in the food industry

Zero-inflated models have been used successfully to describe microbial data with excess of zero-counts. Some sampling plans such as single sampling plan (SSP), resubmitted single sampling plan (RSSP) and quick switching system (QSS) have been developed for the zero-inflated Poisson (ZIP) distribution. Two new plans such as repetitive group sampling (RGS) and multiple dependent state (MDS) sampling are proposed to compare their performance with the existing sampling plans. Additionally, we develop double sampling (DS) and sequential sampling plans for the ZIP data. The plan parameters of the sampling plans are determined by using the unity value approach. The MDS sampling plan is more economical than the other sampling plans, including SSP, RSSP, RGS and QSS, in terms of the required sample size and average sample number in most cases. The probability of a zero in the ZIP data affects the sample size required for all sampling plans. Practical guidance for selecting sampling plans under ZIP is also provided.