Matlab codes for Robust PCA multivariate control chart-Robust PCA multivariate control chart mainly consists two steps:

Step1

Calculates the robust mean and the robust

covariance of original dataset using the

minimum covariance determinant (MCD).

In MCD technique, finding a subset

containing half of the data such that its

covariance matrix has the lowest determinant, then using this subset to

calculate the robust mean and the robust

covariance matrix (Hubert, Rousseeuw, &

Branden, 2005)

Step2

Standardize data using robust mean and

robust standard deviation from Step1.

Apply PCA analysis, calculate the principalcomponent score matrix Y=ZA, where Z is the robust standardized data matrix, and

A is p*p matrix of eigenvectors (also called principalcomponents)