Multivariate Change Point Control Chart Based on Data Depth for Phase I Analysis

Amultivariate change point control chart based on data depth (CPDP) is considered for detecting shifts in either the mean vector, the covariance matrix, or both of the process for Phase I. The proposed chart is preferable from a robustness point of view, has attractive detection performance and can be especially useful in Phase I analysis setting where there is limited information about the underlying process. Comparison results and an illustrative example show that our CPDP chart has great potential for Phase I analysis of multivariate individual observations. The application of CPDP chart is illustrated in a real data example.