Data Mining

17

Multivariate Control Charts

Multivariate control charts monitor multiple variables simultaneously for anomaly detection. This chapter describes three multivariate statistical control charts: Hotelling’s T² control charts, multivariate EWMA control charts, and chi-square control charts. Some applications of multivariate control charts are given with references.

17.1  Hotelling’s T² Control Charts

Let x_i = (x_i1, …, x_ip)′ denote an ith observation of random variables, x_i1, …, x_ip, which follow a multivariate normal distribution (see the probability density function of a multivariate normal distribution in Chapter 16) with the mean vector of μ and the variance–covariance matrix ∑ (see the definition of the variance–covariance matrix ...