A one-sided MEWMA control chart for Poisson-distributed data

This paper introduces a one-sided multivariate exponentially weighted moving average (MEWMA) chart for detecting an increase in a process mean. The control limits are based on the multivariate Poisson distribution and have applications for both industrial processes and public health data. The statistical performance of the proposed MEWMA is examined using run length distributions and is also compared to the traditional MEWMA based on normal-theory limits. Two out-of-control scenarios are of interest: 1) detecting a single point plotting beyond the control limits; 2) a run of two or more points in a row.