Poisson分布截尾序贯近似最优检验的研究——以电子商务产品的抽样检验为例

Study on Truncated Sequential Approximate Optimal Test of Poisson Distribution—Taking the Sampling Inspection of E-Commerce Products as an Example

摘要: 对测量指标服从Poisson分布的电子商务产品进行抽样检验时，为降低抽样检验的试验成本，本文采用样本空间排序法设计Poisson分布计数型截尾序贯近似最优检验方案TS。结果表明，TS方案在保证实际犯两类错误的概率小于给定检验水平的情况下，平均试验次数尽可能达到最小，很好地降低了电子商务产品的抽样检验成本。此外，本文还使用蒙特卡洛法对其准确性进行验证。证实了采用样本空间排序法设计的Poisson分布计数型截尾序贯近似最优检验方案TS具有高可靠性。

Abstract: In order to reduce the test cost of sampling inspection when sampling inspection of e-commerce products whose measurement indexes obey Poisson distribution, this paper designs a sequential approximate optimal inspection scheme TS with Poisson distribution counting type by using the sample space sorting method. The results show that TS scheme can minimize the average number of tests as much as possible under the condition that the probability of actually making two kinds of mistakes is less than the given inspection level, which greatly reduces the sampling inspection cost of e-commerce products. In addition, this paper also uses Monte Carlo method to verify its accuracy. It is proved that TS, a counting truncated sequential approximate optimal test scheme of Poisson distribution designed by sample space sorting method, has high reliability.