The Linear Dependency Structure of Covariance Nonstationary Time Series.
Abstract
The linear dependence, feedback and casuality structure of covariance nonstationary time series is developed. at every instant in time, the amount of linear dependence between time series vectors is expressible as the sum of the amount of feedback from the first time series vector to the second, the amount of feedback from the second time series to the first and the amount of instantaneous feedback. The parametric modeling of multivariate covariance nonstationary time series and the computation of their interdependency structure from the fitted model are also treated. The time series is modeled by a multivariate time varying autoregressive (MVTVAR) model. The fitted MVTVAR model yields an instantaneous power spectral density (IPSD) matrix, The IPSD is used in computing the linear dependency structure of nonstationary time series. An example of the modeling and the determination of instantaneous casuality from a human implanted electrode seizure event EEG is shown. Keywords: Information theory; Time series; Time varying model; Autoregression; Feedback; Casuality; Electroencephalogram.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1987
- Accession Number
- ADA186548
Entities
People
- Will Gersch
Organizations
- Naval Postgraduate School