The ‘effective variance’ control chart for monitoring the dispersion process with missing data

In the last few years, multivariate quality control has been thoroughly studied. The control of the population means vector is usually obtained by using multivariate control charts developed for this purpose. Nevertheless, when the goal is the efficient control of the covariance matrix, few control charts have been developed. In the last few years, the generalised variance control chart, ∣S∣, has been developed to control multivariate dispersion. In processes with a large number of variables to be controlled, it is normal that certain measurements are missing during the control process. For example, in chemical control processes it is commonly found that certain sensors measuring variables online cease to function, producing incomplete data vectors. If this is the case, the generalised variance control chart is of little use. This paper shows the design of the effective variance control chart to control processes with missing data.