Sensitivity and reliability comparisons of EWMA mean control chart based on robust scale estimators under non-normal process: COVID data application

Robust control charts are getting vital importance in statistical process control theory as they are insensitive to the departure from normality. Therefore, the main objective of current work is to determine the effects of non-normal process on the Exponentially Weighted Moving Average (EWMA) control chart. To achieve this goal, the sensitivity and reliability comparisons are made under the non-normal process by comparing five robust M-scale estimators, suggested in literature to modify the EWMA control limits for monitoring process mean and on the basis of percentile bootstrap estimator. The paper addresses the run length (RL) distribution of a robust EWMA control chart under the non-normal process for which the exponential distribution is used as non-normal process. The standard deviation of RL, out-of-control average run length (ARL), and shift detection probabilities are examined to assess the sensitivity and reliability of robust EWMA control charts for mean of monitoring process. The results of this research indicate that the classical EWMA control chart's performance is substantially impacted by the non-normal distribution and the proposed EWMA control charts show higher sensitivity than classical one in terms of having smaller values of out-of-control ARLs. A real-life example from the medical sciences field is provided the practical usage of the proposed control charts. The simulation analysis and practical example have shown that the suggested control charts are effective in quickly monitoring the out-of-control process.