An Alternative Hotelling T^2 Control Chart Based on Minimum Vector Variance (MVV)

Abstract

The performance of traditional Hotelling T 2 control chart using classical estimators in Phase I suffers from masking and swamping effect. To alleviate the problem, robust location and scale estimators are recommended. This paper proposed a robust Hotelling T 2 control chart for individual observations based on minimum vector variance (MVV) estimators as an alternative to the traditional multivariate T 2 control chart for Phase II data. MVV is a new robust estimator which possesses the good properties as in minimum covariance determinant (MCD) with better computational efficiency. Through simulation study, we evaluate the performance of the proposed chart in terms of probability of detection and false alarm rates and compared with the performance of the traditional charts and the chart issued from MCD estimators. The results showed that MVV control chart has competitive performance relative to MCD and traditional control charts even under certain location parameter shifts in Phase I data.