Outlier detection for multivariate time series: A functional data approach

A method for detecting outlier samples in a multivariate time series dataset is proposed. It assumed that an outlying characterized by having been generated from different process than those associated with the rest of series. Each described means estimator its quantile cross-spectral density, which treated as functional datum. Then score assigned to each using depths. broad simulation study shows proposed approach superior alternatives suggested literature and demonstrates consideration data constitutes critical step. The procedure runs linear respect both length number series, quadratic dimensions. Two applications concerning financial ECG signals highlight usefulness technique.