Comparison and classification of stationary multivariate time series

This paper presents procedures to compare and classify stationary multivariate time series. The classification procedure is based on the p-value of a test of hypothesis that is performed for every pair of series under consideration. The test of hypothesis is based on the difference between vector autoregressive parameter estimates of the series. Simulation studies show that the test of hypothesis and the classification procedure perform fairly well for series of reasonable length. (C) 1999 Pattern Recognition Society. Published by Elsevier Science Ltd. All rights reserved.