Authors : S.S. Syed-Yahaya, F.S. Haddad, N.I. Mahat, A. Abdul Rahman and H. Ali
Abstract: It is well known that the traditional Hotelling’s T2 chart is inefficient when outliers are presented in the data. To alleviate the problem, this study proposed three robust Hotelling’s T2 control charts using trimmed estimators to replace the usual mean vector and the covariance matrix in the traditional T2 chart. Trimming was done using modified Mahalanobis distance with median as the location measure and one of the robust scale estimators: MADn, Sn or Tn as the scale measure. These modifications consequently produced three different trimmed estimators. Investigation on the performance of the proposed robust charts, measured in terms of false alarm and probability of detection were conducted via. simulation study. The results were compared to the traditional chart. It was discovered that the proposed control charts performed moderately in terms of controlling false alarm and excellently in terms of probability of detection, surpassing the performance of traditional control chart regardless of conditions.
S.S. Syed-Yahaya, F.S. Haddad, N.I. Mahat, A. Abdul Rahman and H. Ali, 2019. Robust Hotelling’s T2 Charts with Median based Trimmed Estimators. Journal of Engineering and Applied Sciences, 14: 9632-9638.