El efecto de la estimación de la matriz de covarianzas en el desempeño de la carta |s|
Pérez Sánchez, Maryoris Mairen | 2020-07-31
When designing a control chart, it is usual that the parameters should be estimated from a preliminary sample of observations. Numerous studies have shown the effect that the estimation of the parameters can have on the performance of the control charts. It has been found that only for relatively large preliminary samples, the performance of control charts with estimated parameters can be considered satisfactory, in terms of the rate of false alarms that the chart produces in Phase II. In this work we evaluate the effect of the estimation of parameters on the performance of the |S| chart. For this, the number of preliminary samples and the size of subgroups from which the performance of the chart is considered adequate were determined approximately, finally a model is was adjusted that allows the size of the preliminary sample to be related to the false alarm rate observed. For our study, simulation procedures were used to observe the false alarm rate for different sizes of the preliminary sample.