EFFICIENT FITTING OF LINEAR MODELS FOR CONTINUOUS STATIONARY TIME SERIES FROM DISCRETE DATA
Abstract
Consideration is given to the problem of estimating the parameters of the rational spectral density function of a continuous process given n observations equi-spaced in time. The estimates are derived by assuming the observations to be normally distributed. However, it must not be thought that the properties of the estimates depend critically on this assumption. If the dis(over) tribution is non-normal, the estimation process can be thought of as arising from a sort of generalized least-squares method. The joint distribution of the equi-spaced observations is considered, and the relevant parameters of this distribution is estimated first. These are then converted to estimates of the parameters of the spectral density. The estimation of the parameters of the underlying stochastic differentialequation model is also considered. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1960
- Accession Number
- AD0248652
Entities
People
- James Durbin
Organizations
- University of North Carolina at Chapel Hill