ON ESTIMATION METHODS OF INPUT PARAMETERS FOR ECONOMIC ATTRIBUTE CONTROL CHARTS

Applications of the economic design of control charts always assume that all the input parameters are known. Some of these input parameters can be estimated from past observations. In this paper, we consider estimation methods of some input parameters for the economic design of np-charts, namely, the proportions of non-conforming items under both in-control (p0) and out-of-control (p1) states and the rate of occurrence of an assignable cause (lambda) based on two methods: maximum likelihood estimation using the EM algorithm and a naive method. We compare the two methods by using a simulation study. We conclude that the EM algorithm produces better results under general conditions. Thus we recommend the use of the EM algorithm so that the economic np-chart becomes more feasible to use in practice.