A non-parametric control chart for high frequency multivariate data

Abstract

Support Vector Data Description (SVDD) is a machine learning technique used for single class classification and outlier detection. A SVDD based K-chart was first introduced by Sun and Tsung [4]. K-chart provides an attractive alternative to the traditional control charts such as the Hotelling's T2 charts when the distribution of the underlying multivariate data is either non-normal or is unknown. But there are challenges when the K-chart is deployed in practice. The K-chart requires calculating the kernel distance of each new observation but there are no guidelines on how to interpret the kernel distance plot and draw inferences about shifts in process mean or changes in process variation. This limits the application of K-charts in big-data applications such as equipment health monitoring, where observations are generated at a very high frequency. In this scenario, the analyst using the K-chart is inundated with kernel distance results at a very high frequency, generally without any recourse for detecting presence of any assignable causes of variation. We propose a new SVDD based control chart, called a kT chart, which addresses the challenges encountered when using a K-chart for big-data applications. The kT charts can be used to track simultaneously process variation and central tendency.