Answers are grounded in this paper and cite their sources.
In profile monitoring applications, the errors distribution is usually misspecified in practice. The existing monitoring methods are based on conditional mean regressions, which are easily affected by nonnormal or asymmetric error distributions. Thus, a flexible model, the quantile regression model (QRM), is used to model the linear profiles with any distributed errors. The QRM has numerous quantile regressions which can describe more details about linear profiles. Then, a flexible distribution-free exponential weighted moving average type control chart is proposed based on the median regression coefficients and the scale-shift effects of QRM, which could detect the location and shape shifts simultaneously. The design and implementation of the developed chart is given, as well as the estimates of in-control parameters. An illustrative example shows the effectiveness and efficiency of the proposed control chart. The effect of the estimated in-control parameters on the in-control performance is investigated via simulations. Further comparisons show that the proposed chart performs better than the alternative method in most cases for monitoring linear profiles with any distribution errors.
No takes yet. Share an insight, caveat, or question.
Zhang et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: