MULTIPLE TIME SERIES MODELLING.
Abstract
The paper seeks to provide a general framework for the theory and practice of multivariate analysis of time series. It seeks to compare: (1) Spectral approaches to finding relations among time series. (2) Time domain or innovations approaches to finding relations among time series. The paper also seeks to focus attention on: (3) Innovations approaches to cross-spectral estimation. (4) The problem of multivariate analysis of the joint innovations covariance matrix and the sampling properties of its estimators. The various sections are entitled: (2) Innovation Approaches to Modelling; (3) Spectral Approaches to Modelling; (4) Relations Between Time Series; (5) Autoregressive Approach to a Single Series; (6) Multiple Spectral Density Estimation. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 08, 1968
- Accession Number
- AD0672023
Entities
People
- Emanuel Parzen
Organizations
- Stanford University