TIME SERIES ANALYSIS FOR MODELS OF SIGNAL PLUS WHITE NOISE.
Abstract
This paper is concerned with fitting to an observed discrete parameter time series X(.) a model of signal plus white noise. It is shown that there is a basic parameter, representing the signal proportion, to be estimated. Three methods are proposed for estimating the signal proportion, which are named as follows: (1) the cross-bispectral transfer function method, (2) the auto-bispectral transfer function method, (3) the auto-spectral empirical method. A few applications of the model are described.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 12, 1966
- Accession Number
- AD0655829
Entities
People
- Emanuel Parzen
Organizations
- Stanford University