Grouping multivariate time series variables

Abstract

In many industrial and medical applications it is important to identify relationships in multivariate time series (MTS) variables in as short a time as possible. Within this paper, we present a method for decomposing high dimensional MTS into mutually exclusive subsets of variables where within-group dependencies are high and between group dependencies are low. The method involves the use of two evolutionary computation techniques, which find an approximate solution to an otherwise NP-hard problem. We apply the proposed method to two real-world datasets, a chemical process MTS from an oil refinery and an ophthalmic MTS regarding glaucomatous deterioration. Show more