Abstract:Many applications involve monitoring incidence rates of the Poisson distribution when the sample size varies over time. Recently, a couple of cumulative sum and exponentially weighted moving average (EWMA) control charts have been proposed to tackle this problem by taking the varying sample size into consideration. However, we argue that some of these charts, which perform quite well in terms of average run length (ARL), may not be appealing in practice because they have rather unsatisfactory run length distributions. With some charts, the specified in-control (IC) ARL is attained with elevated probabilities of very short and very long runs, as compared with a geometric distribution. This is reflected in a larger run length standard deviation than that of a geometric distribution and an elevated probability of false alarms with short runs, which, in turn, hurt an operator's confidence in valid alarms. Furthermore, with many charts, the IC ARL exhibits considerable variations with different patterns of sample sizes. Under the framework of weighted likelihood ratio test, this article suggests a new EWMA control chart which automatically integrates the varying sample sizes with the EWMA scheme. It is fast to compute, easy to construct, and quite efficient in detecting changes of Poisson rates. Two important features of the proposed method are that the IC run length distribution is similar to that of a geometric distribution and the IC ARL is robust to various patterns of sample size variation. Our simulation results show that the proposed chart is generally more effective and robust compared with existing EWMA charts. A health surveillance example based on mortality data from New Mexico is used to illustrate the implementation of the proposed method. This article has online supplementary materials.

A Statistical Control Chart Formonitoring High-Dimensional Poisson Data Streams

Control chart for monitoring multivariate COM-Poisson attributes

Cumulative sum charts for monitoring the COM-Poisson processes

A Multivariate Control Chart for Monitoring Trivariate Poisson Processes with Spatial Correlation

Two Approaches to Monitoring Multivariate Poisson Counts: Simple and Accurate

Efficient Monitoring of Autocorrelated Poisson Counts

The Use of Probability Limits of COM–Poisson Charts and Their Applications

A Flexible and Generalized Exponentially Weighted Moving Average Control Chart for Count Data

Novel Nonparametric Control Charts for Monitoring Dispersion of Count Data

Joint Monitoring of Mean and Dispersion of Count Data

An Efficient Online Monitoring Method for High-Dimensional Data Streams

Multivariate Exponentially Weighted Moving-Average Chart for Monitoring Poisson Observations

Exponentially Weighted Moving Average Control Charts For Monitoring Increases In Poisson Rate

On-Line Monitoring Data Quality of High-Dimensional Data Streams

Statistical Process Control of Over-Dispersed Multivariate Count Data

Monitoring Poisson Count Data with Probability Control Limits when Sample Sizes Are Time Varying

A Multivariate Non-Parametric Control Chart Based on Run Test.

A Change-Point Based Control Chart for Detecting Sparse Changes in High-Dimensional Heteroscedastic Data

One-Sided Ewma Control Charts for Monitoring Poisson Processes with Varying Sample Sizes

Likelihood-Based Ewma Charts for Monitoring Poisson Count Data with Time-Varying Sample Sizes

A Multivariate Control Chart for Simultaneously Monitoring Process Mean and Variability.