Monitoring process variability using auxiliary information (Q626205)

A construction of Shewhart-type control charts is considered for the changes of variance of a controlled variable \(Y\) in the case when some auxiliary variable \(X\) is also observed. Bivariate normality of \((X,Y)\) is assumed. The chart is based on a regression-type estimate for the variance of \(Y\) by observations on \(X\). The central line, lower and upper control limits for the chart are constructed. The performance of the chart (V-chart) is compared to the performance of the standard \(S^2\)-chart via simulations. The authors conclusion is that the V-chart outshines the \(S^2\)-chart when the correlation between \(X\) and \(Y\) is large enough, e.g.. for sample sizes \(n=100\) it must be greater then 0.2541.